Female Status and Cultural Evolution: A Study in Ethnographer Bias

43
,j. J f J j BEHAVIOR SCIENCE RESEARCH VOLUME 11, NUMBER 3, 1976 Human Relations Files, 2054 Y. S., New Haven, Conn. 06520 Female Status and Cultural Evolution: A Study in Ethnographer Bias William Tulio Divale* The theory that female status will increase with societal complexity is tested. Two independent measures of female status are compared with a measure of societal complexity. Both tests produce significant results, but in a direction opposite that of the hypothesis. With increases in societal complexity, there is a decrease in female status. Four data quality control factors were routinely applied to test for the presence of systematic eth- nographer bias. The DQC tests suggest that the negative relationship be- tween female status and complexity is a spurious one, which can be plausi- bly explained by the ethnographer's sex, length of field stay, and ability to speak the native language, as well as the century in which the fieldwork was done. Further analysis suggests that unbiased reporting of female sta- tus is possible by female ethnographers and by male ethnographers who have worked under optimal field conditions (field stays of over one year, plus native language familiarity). Among this group of observers, the orig- inal hypothesis receives support: high female status is found at all levels of complexity, while low female status is found primarily among less com- plex societies. [Accepted for publication: February 1976.] Introduction This paper is concerned with the relationship between female status and cultural evolution and, consequently, with the reliability of ethnographic reports from which data on these variables are obtained. Conditions of ethnographic *William Tulio Divale (Ph.D., 1974, SUNY at Buffalo) is currently As- sistant Professor and Coordinator of Anthropology at York College of the City University of New York. His interests· include warfare, social struc- ture, and demography; he is also conducting a project in agricultural de- velopment in the Canary Islands. His previous publications include: "Mi- gration, External Warfare, and Matrilocal Residence,'' Behavior Science Research 9 (1974): 75-133, "Temporal Focus and Random Error in Cross- Cultural Hypothesis Tests," Behavior Science Research 10 (1975): 19-36, and""UsTng Date ofEuropean Contact for Time-Lagged Variables in Cross- Cultural Surveys," Behavior Science Research 11 (1976): 39-55. 169

Transcript of Female Status and Cultural Evolution: A Study in Ethnographer Bias

l· ,j. J

f J j

BEHAVIOR SCIENCE RESEARCH VOLUME 11, NUMBER 3, 1976

Human Relations Are~ Files, 2054 Y. S., New Haven, Conn. 06520

Female Status and Cultural Evolution: A Study in Ethnographer Bias

William Tulio Divale*

The theory that female status will increase with societal complexity is tested. Two independent measures of female status are compared with a measure of societal complexity. Both tests produce significant results, but in a direction opposite that of the hypothesis. With increases in societal complexity, there is a decrease in female status. Four data quality control factors were routinely applied to test for the presence of systematic eth­nographer bias. The DQC tests suggest that the negative relationship be­tween female status and complexity is a spurious one, which can be plausi­bly explained by the ethnographer's sex, length of field stay, and ability to speak the native language, as well as the century in which the fieldwork was done. Further analysis suggests that unbiased reporting of female sta­tus is possible by female ethnographers and by male ethnographers who have worked under optimal field conditions (field stays of over one year, plus native language familiarity). Among this group of observers, the orig­inal hypothesis receives support: high female status is found at all levels of complexity, while low female status is found primarily among less com­plex societies. [Accepted for publication: February 1976.]

Introduction

This paper is concerned with the relationship between female status and cultural evolution and, consequently, with the reliability of ethnographic reports from which data on these variables are obtained. Conditions of ethnographic

*William Tulio Divale (Ph.D., 1974, SUNY at Buffalo) is currently As­sistant Professor and Coordinator of Anthropology at York College of the City University of New York. His interests· include warfare, social struc­ture, and demography; he is also conducting a project in agricultural de­velopment in the Canary Islands. His previous publications include: "Mi­gration, External Warfare, and Matrilocal Residence,'' Behavior Science Research 9 (1974): 75-133, "Temporal Focus and Random Error in Cross­Cultural Hypothesis Tests," Behavior Science Research 10 (1975): 19-36, and""UsTng Date ofEuropean Contact for Time-Lagged Variables in Cross­Cultural Surveys," Behavior Science Research 11 (1976): 39-55.

169

170 Behavior Science Research 3, 1976

fieldwork vary, and the abilities of ethnographers vary. It is well known that field reports differ in their reliability and accuracy. In spite of this, perhaps because of a "gentlemen's agreement" among ethnographers, the accuracy of an eth­nographer's field report is rarely questioned in print. Nota­ble exceptions are Redfield's (1930) and Lewis' (1951) dif­fering accounts of Tepoztlan, and Mead's (1938-49) dispute with an exspouse (Fortune 1939) over the temperament of Arapesh males (see Naroll 1970: 928-29 for a discussion of these cases). Another instance is Benedict's interpretation of the Zuni as Apollonian, wherein, among other things, she claimed that "drunkenness is repulsive to them" (Benedict 1934: 82). In contrast, Bamouw (1963: 44) cites several an­thropologists who claim "by far the most common 'crimes' at Zuni are drunkenness and drunken driving." (See also Harris 1968: 404-09 and Pelto 1970: 30-35 for a discussion of the Zuni controversy.)

In spite of the above examples, which indicate that at least one of the parties must be wrong, and in spite of the fact that most fieldwork reports are taken at face value, there is some evidence which suggests that most ethnographic reports are generally very reliable (Driver 1970: 626-27; Naroll 1970: 930-31). Even though most ethnography is re­liable, however, errors do exist; and, more importantly, cer­tain sections or topics in ethnographic reports probably con­tain more error than other topics. A report may be accurate as a whole, but certain sections of it may be inaccurate. The use of data quality control measures is an attempt to find those inaccuracies and to measure their influence (Naroll 1962; 1970).

Two types of error are generally found in ethnographic reports-random and systematic. Random error tends to destroy relationships; it will not produce false correlations (Divale 1975). Systematic bias can produce a false corre­lation-this study presents one such false relationship, which even survived replication. In distinguishing between random and systematic error in cross-cultural tests, Naroll (1962: 12) writes:

Female Status and Cultural Evolution 171

In considering data quality control it is essential to distin­quish two kinds of errors in data reports: random errors and bias. Random errors occur from caJ:elessness or oversight aris­ing from circumstances that make error in either direction equally likely. Biases occur where factors dispose toward error in one direction rather than the other: biases need not necessarily reflect either the conscious or unconscious desires or preconceptions of the observer, for they can also arise out of the observation situation or the phenomenon being ob­served.

To measure the possible influence of systematic error, Naroll (1962; 1970) has developed the control factor meth­od. In this procedure, the theoretically relevant trait is com­pared with a control trait which, it is thought, may possibly be influencing the data collection process (Naroll 1962: 85-105; 1970: 935-38). For example the ethnographer's abil­ity or inability to speak the native language could influence the accuracy of certain data collected. Not being able to speak the native language would probably not influence the accuracy of measurements of house size or construction techniques, but it could influence estimates of the amount of witchcraft practiced (Naroll 1962: 88-89).

Although Naroll has stressed the importance of data qual­ity control, until recently there has been little hard evi­dence showing the necessity for its use. In his original monograph on the subject, Naroll (1962: 84) compared four theoretical variables with eight control factors and found only three out of thirty tests to be significant-not rpany more than would be expected by chance alone. Also, no two substantive variables were significantly related to the same control factor, which would be necessary if a biased corre­lation were to result. Several years lat~r, Schaefer (1973: 220), in a study of family structure and drunkenness, com­pared eighteen theoretical traits with six control factors, but only three correlations were significant. In a recent study, (Divale 1974: 373-74), I found five out of thirty-five quality control tests significant, which is more than would be ex­pected by chance. But as in Naroll's and Schaefer's studies, the same control factors were not significantly correlated

172 Behavior Science Research 3, 1976

with two substantive traits that were theoretically linked. Thus the control trait could not have caused the correlation between the substantive traits. These studies demonstrated that ethnographic reports carry systematic error with re­spect to certain measures, and these studies demonstrated that the control factor method had the ability to articulate systematic error; but the really important issue concerning data quality-the possible production of a spurious correla­tion-was not demonstrated until recently.

In a study of ethnographer bias, Rohner, DeWalt, and Ness (1973) report on two associations which disappeared after the effects of systematic bias were removed and two associations which became statistically significant when bias was removed. The study stems from Rohner's research into parental rejection-acceptance behavior and consequen­tial personality development in children and adults. In their most striking result, Rhoner, DeWalt, and Ness (1973: 293) found an original correlation ofr = .72 (p = .01, N = 11) be­tween parental rejection-acceptance behavior and a child's self-evaluation. After removing the effects of six control fac­tors, the correlation was reduced to r = .13, which was no longer significant. This was the first published instance of a correlation shown to be spuriously produced by ethnogra­pher bias. Their study made a very strong case for the use of data quality control factors in cross-cultural research. But even this instance will fail to convince many sceptics, be­cause Rohner's study involves many hypotheses, and most of them were not influenced by systematic bias. Thus the chance that Rohner's study could have had misleading re­sults overall is small (Rohner, DeWalt, and Ness 1973: 294).

In the present study, a clear case is given for the need for data quality control. The hypothesis that increases in cul­tural evolution will result in an increase in the status of women is tested. Two independent measures of female status are compared with a measure of societal complexity, and both tests are found to produce similar and statistically significant results. However, the direction is opposite that of the hypothesis-increasing societal complexity resulted

Female Status and Cultural Evolution 173

in declining female status. In the eleventh hour, so to speak -the results were almost published as a substantive finding -data quality tests demonstrated that the results of both the original test and the replication were spurious and the result of systematic ethnographer bias. It turns out that the negative relationship between female status and societal complexity can be plausibly explained by the sex of the eth­nographer, his length of field stay, his ability to speak the native language, and the century in which the fieldwork was done.

Cultural Evolution and Female Status

It has long been held by evolutionists that an increase in societal complexity results in a corresponding rise in the status of women, to a point where their status is more equal to the status of men. In The Science of Society Sumner and Keller (1927: 1826) suggested that as society became indus­trial, "woman, carried along into the new phase, made a gain in status."

With increases in societal complexity comes an increase in the size of settlements (urbanization) and an increase in technoeconomic specialization (Naroll 1956). Recent re­search has shown that the maximum settlement population of a society, which is an aspect of urbanization, is probably an excellent index of societal complexity. Many writers, especially for the developing nations of Africa, have linked increasing urbanization, and the economic development which goes along with it, to a rise in the status of women. The underlying assumptions are that in band and tribal so­ciety, females have much lower status than males, and that with increasing societal complexity come opportunities for females to raise their status to a par with males. In his book, African Women in Towns, Kenneth Little (1973: 180) writes:

In fact, urbanization and the women's efforts to alter their po­sition go hand in hand because the growth of urban economic and other institutions automatically holds out new and extra opportunities of achieving status.

174 Behavior Science Research 3, 1976

In a paper entitled: "Urbanization and the Position of Wom­en," Tanya Baker and Mary Bird (1959: 115) concur with Little that "urbanization has in general offered women a greater degree of personal choice and individual activity; and it has offered them personal rewards in personal sta­tus." Similarly a report of a conference held by the Wom­.en's Africa Committee on the Role of Women in Africa (Schuller and Wheeler 1959: 6-7) concluded as follows:

Urbanization and industrialization have opened new avenues for women to participate in many new social institutions. City life has fostered self consciousness and grouped together per­sons with similar interests. It has provided opportunities for education and social mobility. Women are gradually being drawn into the labor force as equals with men.

The connection between urbanization and increasing female status is, of course, the independence which results from the new economic roles opened to women. Marie­Helene Lefaucheux, past president of the International Council of Women, writes (1962: 16):

The many tasks a woman must perform, and above all her role as a mother, may ensure that she receives protection, respect, affection-but never authority. But with economic development this inferior status of the female is declining .... As a general rule women occupy but a minor place in the var­ious social systems, and owe this place entirely to their con­stant toil and their role in the transmission of life.

The way to achieve parity with men, according to these writers, is to integrate women into the economy. Lefau­cheux (1962: 15) writes:

The methods of obtaining equality may differ between coun­tries and between areas of the globe but in the developing countries of Africa surely the best means of ensuring that women have their rightful place is to find ways of including them in the very process of economic development.

The relationships between increasing urbanization, eco­nomic development, female participation in the economy, and a rise in female status may not be as simple and direct as these writers would suggest. For example Knudsen

Female Status and Cultural Evolution 175

(1969: 186-87) has shown that although the participation of women in the U.S. economy has increased, their median income, as a percentage of male income, has decreased. In 1940, women composed 28 percent of the labor force; in 1966 they composed 35 percent-an increase of 7 percent. However, in 1939 the median income for females was 61 percent of the median income for males in comparable jobs. In 1966 the median income for females had declined to 57 percent of the median income for males in comparable jobs. If income is an indication of status, then the status of wom­en in the U.S. declined during the period, even though their numbers in the work force increased. This is similar to Bos­erup's (1970) findings for developing nations. I mention this case only to suggest that the relationship posed by some writers (for example see Benston 1969; Goldberg 1970; Hogeland 1970; and Trey 1972) is probably more complex than they suggest. The original purpose of this paper was to test empirically an assumption which either is actually stated by these writers or underlies their perspectives. This assumption is that increasing societal complexity will result in a higher status for women. To my knowledge, this hy­pothesis was never tested on a worldwide basis, although aspects of it are currently being examined by Margaret Bert­ner (1974) and Maud Naroll (1974).

Sampling

The sample used was taken from Simmons' (1937) study. It is not a random sample; and, not being as such, it repre­sents a weakness of this paper. On the selection of the sam­ple, Simmons (1937: 495) writes:

Our data are derived ... from an intensive study of the litera­ture on seventy-one specific tribes and peoples selected with due regard to climatic zones, geographical distribution, racial differences, and culture areas. Sixteen tribes are chosen from native North America, ten from Central and South America, fourteen from Africa, three from Europe, sixteen from Asia, and twelve from Oceania and Australia ... The total culture of each tribe was studied, but a special effort was made to secure classifiable information on 109 subjects of inquiry per-

176 Behavior Science Research 3, 1976

taining to maintenance activities, political and social organi­zation, and religious beliefs and practices.

Simmons' sample was used because he had already coded female status, and this provided a second and independent measure of status. I estimated that the risks involved in not selecting a random sample were sufficiently offset by the advantage of having a second measure of female status.

Information on maximum settlement size (my index of cultural evolution) was available for only thirty-one of the seventy-one societies. Simmons measured female status in twenty-nine of the thirty-one, and I was able to find status estimates for all thirty-one. This greatly reduces the size of the original sample. While larger samples are generally more reliable, it is also easier to achieve statistical signifi­cance in larger samples. Thus significant results that are found in spite of small sample sizes should not be dis­counted.

The range of societal complexity in the sample runs from very simple societies, such as the Kung Bushmen and the Tasmanians, to complex, preindustrial empires, such as the Aztecs and Inca. No modern industrial states are included; and, as a result, it would be hazardous to generalize the findings to modern states.

Simmons' Measure of Female Status

For his seventy-one-society sample, Simmons measured 109 variables on the same four-point ordinal scale. Simmons (1937: 497) defined the scale as:· (1) "absence or non-appear­ance of the trait, when definitely indicated in the sources;" (2) "incipient presence, slight elaboration or cultural unim­portance of the trait;" (3) "presence without dominance, moderate elaboration, or intermediate importance of the trait;" and (4) "dominance, marked elaboration, or strong social importance of the trait in the particular tribe under investigation." One of Simmons' 109 variables (No. 74), "Subjection or inferiority of women," is a direct measure of female status. For analysis, Simmons' scale was collapsed

Female Status and Cultural Evolution 177

to a two-point ordinal scale. His codes "0" (1) and "-" (2) were coded 1; and, his codes "=" (3) and "+" (4) were coded 2. I did this because his scale was too fine for the na­ture of his data-there was no significant difference be­tween the societal complexity codes of his first two groups and his last two groups. However, when the combined groups were compared with each other, there was a signifi­cant difference between them.

A major weakness of Simmons' measure of female status is that we do not know exactly how this variable was mea­sured. He gives no coding description of the variable other than its name and his codings for each society in the sample. To compensate for this shortcoming, I attempted indepen­dently to measure female status for the societies in the sam­ple.

Author's Measure of Female Status

The difficulty in measuring female status is well known (Sanday 1973; Schlegel 1972; for review essays, see Hochs­child 1973; Evans-Pritchard 1965; Westermarck 1905). But some attempt must be made if we ever hope to achieve understanding of this topic. On the concept of status, Knud­sen (1969: 184) says that "status, though ambiguous in so­ciological usage, refers to relative prestige, esteem, power or recognition of two or more categories within the same general classification." And Laswell (1965: 42-43) says: "In all definitions and uses of the term 'status,' factual compar­ison is implicit." Therefore, female status should always be viewed relative to male status.

In a recent paper Sanday (1973: 1682) took a different approach; she defined female status "in terms of the num­ber of economic and political rights which accrue to wom­en," and she added that "prestige and position are not to be included in this discussion." However, it appears that the relationship between female status and economic and po­litical power, if it exists, is not a simple one. Sanday (1973: 1694) constructed a Guttman scale of female status, which was derived from her definition. The items she included

178 Behavior Science Research 3, 1976

were: female control over produce; demand for female pro­duce; female participation in political activities; and the presence of female solidarity groups. But her major hypoth­esis-in which this scale was used-failed to receive sup­port. When she tested this scale against the importance of female contribution to subsistence, the results were in the opposite direction. Female status was lower when females contributed more to subsistence (r = -.16). In fact, the geo­graphical region in which the society was located turned out to be a better predictor of female contribution to subsis­tence than Sanday's measure of female status.

I have taken a different approach. Like Sanday' s, my measure of female status is objective; but it is based on a subjective assessment, made by the ethnographer of the so­ciety. Status is a subjective concept. It is real, but it is hard to pin down. Thus I sympathize with attempts like San day's to measure status in concrete terms. However, it seems to me that it is extremely difficult to impose an objective framework on an inherently subjective concept, such as status. A more limited but more immediately fruitful ap­proach, perhaps, is to try to measure objectively the subjec­tive concept itself. I have done this by taking the ethnog­rapher's assessment of the overall prestige, esteem, power, or recognition given to the "aven'1.ge" woman in comparison to the "average" man in a society. Granted, such an ap­proach is open to substantial amounts of bias-introduced either by myself or through the ethnographic accounts sur­veyed.

To control for any systematic bias that I could have intro­duced myself, coding decisions were made by two female coders who were not aware of the hypothesis. They could not possibly have introduced bias, since they did not know the hypothesis. This is not to say that my measure of female status is error-free; as we shall see, it most certainly con­tains error. However, the error that is present has not been introduced by myself, since naive coders were used. Any error present is either random-which tends to destroy cor­relations (Divale 1975)-or it is systematic bias introduced

Female Status and Cultural Evolution 179

by ethnographers-which can produce a spurious correla­tion (Blalock 1972: 413-14). In comparison to Simmons' measure of status, it is known exactly how this variable was measured. The actual statements of ethnographers regard­ing female status, including page and source references, are listed in the Appendix. Thus the reader can decide for himself as to the validity and reliability of the coding proce­dures.

Cultural Evolution

The size of the largest settlement population in a society is used as a measurement of that society's level of cultural evolution. An extensive discussion of cultural evolution and why maximum settlement population is its best all-around index is given in Naroll and Divale (1976) and in Divale (1974); the following is a summary of that discussion. In 1956, Naroll proposed an index of societal development which was composed of three measures: maximum settle­ment size, occupational specialization, and organizational ramification. Other measures of societal complexity, using different indexes, were later proposed by Freeman (1957; Freeman and Winch 1957), Carneiro (1970; Carneiro and Tobias 1963), Bowden (1969), and Murdock and Provost (1973).

Tatje and Naroll (1970) compared Naroll's (1956) index with Freeman's (1957) index, and a rank-order correlation of .89 was fo"Und. Schaefer (1969) then compared Marsh's (1967) index with the indexes of Naroll and Freeman. Ken­dall's coefficient of concordance between these three scales was .92. Murdock and Provost (1973) presented a ten-item scale of cultural complexity, which incorporates the. two variables used in Marsh's (1967) index for preliterate soci­eties; and the Murdock and Provost scale is highly correlated (rho = .94) with Carneiro's scale. Recent research by Bow­den (1972) indicates that his "index of sociocultural devel­opment"-which is derived from Simmons' (1937) data and correlates highly with Carneiro's (1970) scale-is correlated at .97 with the logarithm of the maximum settlement popu-

180 Behavior Science Research 3, 1976

lation-one of the variables in Naroll's index (see Otterbein 1972: 26-27, for a discussion of settlement size). This re­search by Bowden suggests that the single variable of the maximum settlement population of a society is, by itself, an excellent measure of societal complexity-it measures cultural complexity as well as Carneiro's scale of 618 traits, which is prohibitively expensive in research time to com­pute.

It is clear from the high intercorrelations that these sever­al measures of cultural evolution are all measuring the same thing-although each from a slightly different approach. It is also fairly certain that cultural evolution itself is being measured, because Naroll's index taps three of the most im­portant aspects of culture-urbanization, economic speciali­zation, and social organization-and Carneiro's measure is an exhaustive Guttman scale of 618 cultural traits.

To measure maximum settlement size, the coders-who were not aware that this variable represented societal com­plexity-examined the ethnographic sources for census re­sults or population estimates. The population estimates were then ranked into a nine-point scale. In contrast to the female status variable, this measure is almost entirely ob­jective. For example Hitchcock (1891: 38-439) says that the Ainu village of Hidaka contained 1,669 persons in 1888 and was the largest population in Ainu society. Hitchock's esti­mate could be wrong, but it is probably not off by much. In other words, this variable is probably fairly accurate; and it certainly is objective. In the appendix, the maximum settle­ment size for each society is given, along with the name of the settlement and the time period of the census. Page and source references are also provided; the reader may check the validity of these measurements for himself.

Test Results

Analysis was conducted through the aid of the SPSS com­puter program (Nie et al. 1975). The coefficients of correla­tion computed were gamma and tau-c, which are suited to ordinal level variables. Gamma makes no corrections for

Female Status and Cultural Evolution 181

ties or table sizes. Tau-c was chosen over the more frequent­ly used tau-b, because in addition to making corrections for ties, tau-c is more suited to the rectangular tables produced by my measures of status and complexity. Tau-b is better suited to square contingency tables (where the number of rows and columns are equal). Formulas for these measures are in Nie et al. (1975: 227-28).

In Table 1 the hypothesis is tested, using Simmons' mea­sure of female status. The correlation is moderate (gamma = .35), but it is statistically significant (p = .01, N = 29). While a significant relationship has been found, it is in the oppo­site direction of the hypothesis. The theory held that female status should increase-become on a par with male status­with increases in societal complexity. However, just the op­posite occurs-female status declines as societal complexity increases. In Table 2 the hypothesis is retested, using my measure of female status; again similar and significant re­sults are found (gamma = .30, p = .01, N = 31). The direc­tion is also opposite of the hypothesis: female status de­clines as societal complexity increases.

Although we do not know how Simmons' variable was measured, we do know how my variable was measured, and there is a high degree of agreement between the two mea­sures (gamma = .79, p = .0001, N = 29). Since both tests reach similar results, the hypothesis that female status in­creases with societal complexity becomes doubtful. How­ever, in the course of the analysis some findings emerged which question the validity of the previous tests. This is one of the clearest cases of a plausible relationship that is prob­ably due entirely to systematic reporting errors of ethnog­raphers. These findings, to be discussed below, validate completely the necessity of using quality control factors in comparative research (Naroll 1962; 1970).

Ethnographer Bias

In setting up the research design of this study, I con­structed several data quality control variables to assess the presence of systematic errors in the data. I did this quite

182 Behavior Science Research

Table 1. Cultural Evolution and Simrrwns' Measure of Female Status*

Size of Maximum "Subjection or Inferiority Settlement of Women

Number of Absent or Present or Persons Incipient Elaborate

48--60 Bushmen Semang Tasmanians

I ban 134-196 Jivaro

Lapps Maori

Andamanese Porno

215-299 Seri Yahgan Chukchee

Hottentot 308-420 Trobriands

Yedda Chippewa

Navaho 507-720 Shilluk

Tuareg Toda

956-1,669 Omaha Samoans Ainu

3,000-6,400 Iroquois Fang

35,000-70,705 Ashanti Albanians Araucanians

200,000-400,000 Aztecs

Totals U9 10

3, 1976

Total

3

4

5

4

4

3

2

3

1

29

gamma = .35 Kendall's tau-c = .29 Probability tor tau-c .012 Note: see Appendix for codings of individual societies. * From Simmons 1937.

Female Status and Cultural Evolution 183

Table 2.

Size of Maximum Settlement

Number of Persons

48-60

134-196

215-299

308--420

507-720

956-1,669

3, Q00-6, 400

35,000-70,705

200,000--400,000

Totals

Cultural Evolution and Divale's Measure of Female Status

Ethnographer's ] udgment of Female Status

Equal to Males or

Slightly Below

Semang

I ban Jivaro Lapps

Andamanese Porno Seri Yahgan

Hottentot Trobriands

Navaho Shilluk Toda Tuareg

Ainu Omaha Samoans

Albanians

18

Markedly Below Males

Bushmen Tasmanians

Maori

Chukchee

Chippewa Vedda

Crow Fang Iroquois

Araucanians Ashanti

Aztecs Inca

13

gamma= .30 Kendalfs tau-c = .27 Probability for tau-c = .013 Note: see Appendix for codings of individual societies.

Total

3

4

5

4

4

3

3

3

2

31

184 Behavior Science Research 3, 1976

matter-of-factly and did not expect to find evidence of signif­icant bias; in fact, I hoped not to find evidence of bias. I used the controls simply in the interest of scientific rigor and planned to report, in a line or two, that controls were used but evidence of bias was not found. I constructed four control variables, which could have influenced the substan­tive measures. I was particularly concerned about the accura­cy of the female status variable, but the controls were appli­cable to the societal complexity measure as well. The four control factors are: the sex of the ethnographer, the length of the ethnographer's field stay, the ethnographer's ability to speak the native language, and a fourth measure-which I called the Victorian effect-the century in which the field­work was done.

The sex of the ethnographer is an obvious source of po­tential bias concerning female status-especially since the variable was coded on the basis of ethnographers' state­ments. It might be expected, if bias were present, that fe­male ethnographers would tend to rate female status arti­ficially high, while male ethnographers would tend to rate it artificially low. If there was no significant difference be­tween them in their ratings of female status, then it could be assumed that the observer's sex was not biasing his or her assessment. The overwhelming majority of ethnogra­ers are males-only five of the thirty-one societies had fe­male observers. Since there were so few female ethnogra­phers, if at least one of the sourc~s consulted was written by a female, the society was considered to have been covered by a female ethnographer.

The length of the ethnographer's field stay could also be a source of bias. The longer the field stay, the more familiar the observer would be with the culture; and, presumably, the more accurate would be his reports. Naroll (1962; 1970: 927-73) has found that one year is a delimiting period of time. Ethnographers who spend one year or more in the field should tend to give more accurate reports, on the aver­age, than ethnographers who spend less than one year. If there is no significant difference between estimates of fe-

Female Status and Cultural Evolution 185

male status and settlement size for these two groups, then it may be assumed that the reports of both groups are equal­ly reliable. Sixteen societies were covered by ethnographers who met the one year or more requirement, and fifteen so­cieties were not.

An ethnographer's ability to speak the language of the people he is studying could also be a source of bias. A bet­ter understanding of a culture can probably be had when the observer can speak the native language. This ability would be especially relevant in making estimates of female status, which are subjective and which are credible only when they are made by an objective observer who has a deep understanding of the culture. Fourteen societies had ethnographers who claimed they could speak the native language or who provided evidence of linguistic compe­tence, such as the translation of texts or the publication of grammars; seventeen societies did not have ethnographers who could meet this requirement.

The final control factor is what I call the "Victorian effect." This refers to the century in which the field research was conducted-before 1900 or after 1900. It would seem that ethnographers from a Western cultural tradition-which form the overwhelming majority of observers-would have had a different opinion of what female status should be be­fore 1900 than would ethnographers who conducted their field observations after 1900. Their opinion of what female status should be, or what the actual status of women in their own cultures was, could have biased their assessments of female status in the cultures they studied. OfThand, it might be expected that observers who did their fieldwork before 1900 might systematically tend to report female status as low, while observers who did their fieldwork after 1900 might systematically tend to report it as high when com­pared to the pre-1900 group. If no significant differences were found in the female status ratings of these groups, then it might be assumed that the Victorian effect was not biasing the measurement of female status.

186 Behavior Science Research 3, 1976

Data Quality T~st Results

My measures of female status and societal complexity were compared with each of the control factors, and the tests between female status and societal complexity were rerun, controlling for each of the data quality control fac­tors. In Table 3 the correlations between female status and the four controls are listed; three of the tests are statistically significant, and all of the correlations are moderate to high. Male ethnographers tend to report low female status, while female ethnographers tend to report high female status (gamma = .55, p = .055, N = 31). This means that the sex of the ethnographer is clearly biasing his or her assessment of the status of females in the society being observed. Eth­nographers who spent less than one year in the field tend to report low female status, while ethnographers who spent over a year in the field tend to report high female status (gamma = .19, p = .224, N = 31). The results are not signif­icant at the .05 level, the usual cut-off point in social sci­ence research, but the correlation is still moderate in size. If the sample were larger, say an N of 50, a gamma of .19 would be statistically significant. Additional evidence of bias comes from the next finding-that ethnographers who cannot speak the native language tend to report low female status, while those who can tend to report high female status (gamma = .68, p = .001, N = 31). If we can assume that a more favorable field situation is where the observer spends at least one year and can converse in the native language, then this implies that estimates of low female status are the produce of poor-quality fieldwork. Staying in the field for more than a year and conversing in the native language may generally mean more accurate reporting. However, in the case of female status, this assumption might not apply. It is possible that, just as short field trips and an inability to speak the native language lead to systematically rating fe­male status as low, staying over a year and learning the lan­guage may lead to systematically reporting female status as high. This could be the result of a "leave-my-people-alone" effect, where the ethnographer has grown to care for and

Female Status and Cultural Evolution 187

identify with "his people," and he is hesitant to report any of their activities in a bad light. I do not know whether the above effect applies in this instance, but it is a possibility (see Rohner, DeWalt, and Ness 1973 for a similar instance).

The fourth control variable also showed a significant bias with female status. Ethnographers who conducted their field research before 1900 tended to report female status as low, while ethnographers who did their field research after 1900 tended to report female status as high (gamma = -.64, p = .001, N = 31). Thus ethnographers' reports of female status are being influenced by the Victorian effect. Overall, it appears that the female status measure is heavily in­fluenced by systematic ethnographer bias, as three of the four control factors are significantly associated with it, and the fourth is almost so.

In Table 3 the correlations between cultural evolution and the control factors are also listed. Reports of maximum settlement size seem to be moderately influenced by the sex of the ethnographer (gamma = .-21, p = .197, N = 31). There may be some tendency for females to work in more complex societies, perhaps for reasons of safety or because they are with their families, but the results are not statis­tically significant and could be due to chance. Maximum settlement size does not appear to be influenced by the Vic­torian effect (gamma = -.09, p = .304, N = 31). These two results are consistent, because there does not seem to be a plausible reason why an ethnographer's sex, or whether or not the research was done before or after 1900 should have any bearing on reports of maximum settlement size.

Length of field stay is correlated with cultural evolution, however, as ethnographers who spend less than one year in the field tend to report larger settlement sizes than do those who spend more than a year (gamma = .57, p = .0005, N = 31). I am not sure why this occurs, but the error factor could be that short field stays result in inflated estimates of maximum settlement size, which would be corrected with a longer field stay that might allow time for travel to several communities in the society.

188 Behavior Science Research 3, 1976

Ethnographers who do not speak the native language also tend to report larger settlement sizes than do ethnographers who can speak the language (gamma = .54, p = .001, N = 31). These two findings suggest that "poorer" fieldworkers -if field stays of less than one year and an inability to speak the language can be taken as indexes of poor fieldwork­tend to overestimate the true maximum settlement size of a society. Another possibility is that those ethnographers who stay over a year in the field and learn the native language tend to seek out smaller and less complex societies, which have, of course, smaller maximum settlement populations.

In any case, it is clear from the results in Table 3 that sig­nificant amounts of ethnographer bias have influenced re­ports of both female status and societal complexity. If only one of the variables was significantly associated with the control factors, then the hypothesis test would be unaf­fected, since both variables have to contain bias for a spu-

Table 3. Data Quality Measures with Divale's Measure of Female Status and Cultural Evolution

Data Quality Control Measures Female Status Cultural Evolution

gamma p N gamma p N

Length of Field Stay .19 .224 31 .57 .0005 31 Native Language Familiarity .68 .001 31 .54 .001 31 Sex of Ethnographer .55 .055 31 -.21 .197 31 Victorian Effect -.64 .001 31 -.09 .304 31

(Century of Fieldwork)

Direction of scales: Female Status: 1-equal to males; 2-markedly below males. Cultural Evolution: From 1 through 9, low to high complexity (see Table 1 for specific breakdown). Length of Field Stay: l--one year or more; 2-less than one year. Native Language Familiarity: 1-spoke the language; 2-could not speak the language. Sex of Ethnographers: 1-female or both; 2-male. Victorian Effect: 1-field research conducted before 1900; 2-after 1900.

Female Status and Cultural Evolution 189

rious correlation to be produced. However, since both vari­ables contain evidence of systematic bias, a rival hypothesis becomes plausible-that the results are merely due to sys­tematic reporting errors on the part of ethnographers.

To see if this was the case, the tests between female sta­tus and societal complexity were rerun, although this time controlling for the data quality factors-these tests are re­ported in Table 4. To control for the sex of the ethnog­grapher, for example, the sample was divided into two sub­samples-the five societies that were covered by female ethnographers and the twenty-six societies that were cov­ered by male ethnographers. Female status was then com­pared with societal complexity in each of the subsamples. If the sex of the ethnographer was not influencing the relationship between female status and societal complexity, then gamma should be similar in each of the subsamples. As can be seen in Table 4, this is not always the case, as some­times the two gammas are at opposite poles.

Ethnographers who spend less than one year in the field tend to report declining female status with increasing soci­etal complexity (gamma = .41, p = .014, N = 15); ethnogra­phers who spend more than one year in the field tend to re­port a similar relationship (gamma = .23, p = .139, N = 16), although the correlation is lower. When the Victorian effect is controlled, the results are also similar, whether the field­work was conducted before or after 1900. Ethnographers who made their field observations after 1900 tend to report declining female status with increasing societal complexity (gamma = .47, p = .029, N = 16), and ethnographers who made their observations before 1900 report a similar relation­ship (gamma = .18, p = .186, N = 15), although the corre­lation is lower.

However, completely opposite results are found when language familiarity or sex of the ethnographer is consid­ered. Male ethnographers tend to report declining female status with increasing societal complexity (gamma = .53, p = .0001, N = 26), while female ethnographers tend to re­port increasing female status with increasing societal com-

190 Behavior Science Research 3, 1976

plexity (gamma = -1.0, p = .172, N = 5). Ethnographers who could not speak the native language also tend to report declining female status with increasing societal complexity (gamma = .35, p = .02, N = 17), while ethnographers who could speak the language tend to report increasing status with increasing complexity (gamma = - .23, p = .25, N = 14). The outcome of these tests is clearly dependent either on the sex of the ethnographers or on whether they could speak the native language. The fact only two of the gammas are statistically significant is not important-only four of the eight tests in Table 4 are significant. This is because the sample, which was small to begin with, was divided into evert smaller subsamples; in these circumstances, only the highest of correlations could be statistically significant. The important thing in Table 4 is the magnitude of the gammas, and they are all fairly high-the average for all eight is .43.

Discussion

These results strongly suggest that the associations found in the tests of the original hypothesis (Tables 1 and 2) be­tween declining female status and increasing societal com­plexity are spurious correlations, which are due to systematic reporting errors resulting from nonoptimal field conditions and/or sex bias. It is hard to assess whether the spurious association is primarily the result of poor field conditions or of sex bias-the sample is not large enough to obtain con­clusive results. However, further analysis of these data seems to suggest that reliable reports of sex status can be ob­tained from either female ethnographers or male ethnog­raphers if they spend over a year in the field and can speak the native language.

This conclusion was reached, in part, from the data in Table 5, which shows the varying frequencies of reported levels of female status. For the entire sample, eighteen societies (58 percent) were reported to have high female status, while thirteen (42 percent) were reported to have low female status. However, when the sample is separated

Tab

le 4

. C

ultu

ral

Evo

luti

on a

nd D

ival

e's

Mea

sure

of F

emal

e St

atus

C

ontr

olli

ng fo

r D

ata

Qua

lity

Mea

sure

s

Incr

easi

ng

Incr

easi

ng

Cul

tura

l E

volu

tion

C

ultu

ral

Evo

luti

on

wit

h w

ith

Qua

lity

Con

trol

Var

iabl

es

Con

trol

ling

for

D

eclin

ing

Stat

us

Dec

lini

ng S

tatu

s C

ontr

olli

ng fo

r

gam

ma

p N

ga

mm

a p

N

Len

gth

of

Fie

ld S

tay

One

Yea

r o

r M

ore

.23

.139

16

.4

1 .0

14

15

Les

s T

han

On

e Y

ear

Nat

ive

Lan

guag

e F

amil

iari

ty

Spo

ke L

angu

age

-.2

3

.255

14

.3

5 .0

23

17

Cou

ld N

ot S

peak

Sex

of E

thno

grap

her

Fem

ales

or

Bot

h -1

.00

.1

72

5 .5

3 .0

001

26

Mal

es O

nly

Vic

tori

an E

ffec

t B

efor

e 19

00

.18

.186

15

.4

7 .0

29

16

Aft

er 1

900

Not

e:

Zer

o-or

der

corr

elat

ion

betw

e.en

inc

reas

ing

cult

ural

evo

luti

on a

nd

dec

lini

ng f

emal

e st

atus

is:

gam

ma

=

.30,

p

=

.013

, N

=

31.

If S

imm

ons'

mea

sure

is u

sed,

gam

ma=

.35

, p

= .

012,

N =

29.

A n

egat

ive

corr

elat

ion

mea

ns t

han

incr

ease

s in

com

­pl

exit

y ar

e as

soci

ated

wit

h in

crea

sing

fem

ale

stat

us;

a po

siti

ve c

orre

lati

on m

eans

tha

t in

crea

sing

com

plex

ity

is a

s­so

ciat

ed w

ith

decl

inin

g fe

mal

e st

atus

.

~

~ >=I .....

~

tr.J ~ E' "" >=

I ;:3

$;

)...

(") .: f ..... t-t:l

c 0 E"' .... .... 0 ;:3

1-'

co

1

-'

192 Behavior Science Research 3, 1976

according to the ethnographer's sex, entirely different re­sults are obtained: four out of five female ethnographers (80 percent) reported high female status, while only four­teen out of twenty-six male ethnographers (54 percent) re­ported high female status. At least one, and perhaps both, of these groups is incorrect concerning the frequency of so­cieties that have high female status. On the basis of sex alone, it is not possible to decide which sex is giving more accurate reports. However, if length of field stay and na­tive language familiarity are considered, it would be pos­sible on logical grounds to infer which group is more prob­ably accurate (Witkowski 1975; Naroll 1962). With respect to these variables, a more optimal field situation should occur when the ethnographer spends at least one year in the field and is able to speak the native language; con­versely, a nonoptimal field situation should occur when an ethnographer does not meet these standards. On purely logical grounds it may be assumed that ethnographers work­ing under optimal field conditions would, on the average, have a more thorough knowledge of the culture than those working under nonoptimal conditions. If systematic dif­ferences in reports of observed behavior occur between these two groups, it can be inferred that the reports of the fieldworkers who make their observations under optimal field conditions are probably more reliable than those made under nonoptimal conditions.

Systematic differences in the reporting of female status for optimal and nonoptimal fiefd conditions occur in this sample (see Table 5). Nine out of twelve ethnographers (75 percent) who worked under optimal field conditions report female status as high, while only nine out of nine­teen ethnographers (47 percent) who worked under non­optimal field conditions report female status as high. If we could preclude the possibility of sampling error, we would be more likely to accept as accurate the reports of the field­workers who observed under optimal conditions. The twelve optimal condition fieldworkers (who include eleven males and one female) report high female status about as

Female Status and Cultural Evolution 193

often as do female ethnographers (75 percent and 80 per­cent). Even if only males are considered, the results are the same: eight out of eleven male observers (73 percent) who worked under optimal field conditions report female status as high, while only six out of fifteen males (40 per­cent) who worked under nonoptimal conditions report fe­male status as high. Although the small frequencies pre­clude conclusive results, it may be inferred that more accurate reports of female status can be had from female ethnographers and from male ethnographers if they worked under optimal field conditions (field stay of a year or more and an ability to speak the native language).

Table 5. Frequency of Reported Levels of Female Status by Ethnographers Sex, Length of Field Stay, and

Native Language Familiarity

Reported Female Status as:

Equal to Males or

Slightly Below Markedly Below

Males

Ethnographer Frequency Percent Frequency Percent Total

Entire Sample Females or Females

and Males Males Only Spent One Year in

Field and Spoke Native Language

Spent Less than One Year in the Field and Did Not Speak Native Language

Males Who Spent One Year in the Field and Spoke Language

Males Who Speht Less Than One Year in the Field and Did Not Speak Native Language

18

4 14

9

9

8

6

(58)

(80) (54)

(75)

(47)

(73)

(40)

13

1 12

3

10

3

9

(42)

(20) (46)

(25)

(53)

(27)

(60)

31

5 26

12

19

11

15

194 Behavior Science Research 3, 1976

On the basis of these results, a retest of the hypothesis is in order. Only this time, the sample will be limited to soci­eties observed by female ethnographers or by male ethnog­raphers who have spent at least one year in the field and who speak the native language. A sample so stratified should be without systematic ethnographer bias with re­spect to female status. The results of this test are reported in Table 6. There were sixteen societies in which these condi­tions could be met (five were observed by females and eleven by males who worked under optimal field condi­tions). The results are indeed significant (gamma = -.51, p = .03, N = 16), but the negative gamma means that the re­sults are the opposite of the results in Tables 1 and 2. Socie­ties observed by females or by males who worked under op­timal field conditions indicate that female status is equal to male status or only slightly below as societal complexity in­creases. Looking at the distribution in Table 6, high female status seems to occur in societies at all levels of complexity, but low female status seems to occur only among less com­plex societies. These results appear to support the original hypothesis: that female status improves as society becomes more complex. These results must remain inconclusive, of course, until they can be replicated on another and larger sample.

One interesting implication of these findings is that for judging estimates of female status, females, whether or not they have worked under optimal field conditions, tend to give reliable reports of female status, while only males who have worked under optimal field conditions give status estimates that are as reliable. This is not the first time that this has occurred. In another instance, Naroll, Naroll, and Howard (1961) have shown that female ethnographers are more accurate than male ethnographers in their reports of the position women take during childbirth-upright versus supine. They were testing Howard's hypothesis that the upright position during parturition is more advan­tageous and natural, since the direction of gravity and the direction of the expulsive force are synergized. Howard

Female Status and Cultural Evolution 195

Table 6. Cultural Evolution and Divale' s Measure Female Status

(Sample Includes Only Those Societies Observed by Female Ethnog-raphers, or by Male Ethnographers Who Spent at Least One Year in the Field and Spoke the Native Language.)

Size of Maximum Ethnographer's Judgment Settlement of Female Status

Equal to Males or Markedly Below

Persons Slightly Below Males Total

48--60 Semang Bushmen 2

I ban Maori 134-196 Jivaro 3

Andamanese Chukchee 215-299 Porno 4

Yahgan

308-420 Trobriands Chippewa 2

507-720 Navaho 1

Ainu 956--1,669 Omaha 3

Samoans

35,000-70,705 Albanians 1

Total 12 4 16

gamma= -.51 Kendall's tau-c = -.33 Probability fur tau-c = :029 Note: see Appendix for codings of individual societies.

196 Behavior Science Research 3, 1976

maintained that the supine position-used by most Western physicians-is abnormal and unphysiologic. To test this hypothesis, they examined ethnographic reports in a sample of seventy-six societies for the position of women during childbirth; most societies use some form of the upright position. They also applied control factors to check for eth­nographer bias-one of which was sex of the ethnographer. They found a significant difference in reporting-female ethnographers tend to report the upright position in a ratio of almost six to one, while male ethnographers tend to re­port the upright position in a ratio of only four to one. Be­cause the special nature of the variable suggested that fe­males would have acces to more reliable data, Naroll, Naroll, and Howard concluded that the female ethnographers in their sample gave more accurate reports. Their reasoning was as follows (1961: 951):

A third factor importantly affecting rapport where the study of childbirth is concerned is the sex of the ethnographer. Childbirth of course always involves the intimate affairs of women. In many societies, it is not men's business, and men ordinarily take no part in it. But it generally is a topic which women often discuss among themselves. Presumably, then, a woman ethnographer is likely to have better rapport with women informants and hence less likely to be misled through informants' embarrassment.

The point I wish to make, of course, is that the inaccuracies in reporting that are due to the ethnographer's sex are sim­ply a function ofthe sensitive nature of some variables. Re­porting errors that are a function of inadequate knowledge of the culture can be corrected with longer field stays and by learning the native language, as it appears to have per­haps been the case with some male ethnographers in the present sample. With respect to the present study, though, it must be emphasized that it has not been clearly estab­lished whether the bias in the female status reports is due to both sex bias and nonoptimal fieldwork conditions, or primarily. to sex bias. I was· able to show a difference be- . tween good and poor male ethnographers, but the sample

/

Female Status and Cultural Evolution 197

was too small to compare the female ethnographers. A more definitive answer to the question can only be settled if (1) the results in Table 6 are replicated, and (2) a different re­search design is also used. Raoul Naroll has suggested that it would be necessary to hold the society constant and to vary the ethnographers. A requirement of the research de­sign would be for the society to have been studied by many ethnographers-twenty to thirty. Naive coders would mea­sure female status, length of field stay, and native language ability for the sample of ethnographers; and the sample would be divided by sex. It would then be possible to ascer­tain more conclusively which factor was causing the bias. There are not many societies where these conditions can be met, but they exist-for example among the Navaho, the Hopi, and some Plains Indians societies. This research is planned for the near future.

Acknowledgments

The initial stages of this research were supported by an Ogden Mills Fellowship at the American Museum of Nat­ural History. I would like to thank my research assistants, Debbie Velinsky and Tamar Gordon, for doing the coding. An earlier version of this paper was read in November 1974, at the 73d Annual M~eting .of the American Anthro­pological Association in Mexico City. Helpful comments and criticisms were made by Raoul Naroll, Keith Otterbein, Pertti Pelto, and two anonymous reviewers. I would es­pecially like to thank my friend Stan Witkowski for his very thoughtful and helpful comments.

198 Behavior Science Research 3, 1976

APPENDIX

Codes for Female Status and Cultural Evolution

The thirty-one societies in the sample are listed below in al­phabetical order. The following data are provided: Simmons' code of female status, the author's measure of female status, the size and name of the largest settlement population, and, when avail­able, the ·ethnographer's length of field stay and ability to speak the native language. Page and source references are from the Human Relations Area Files. For complete bibliographic infor­mation, the reader should consult the HRAF Source Bibliography (HRAF 1976). For example in measuring female status for the Ainu, I have quoted a passage from Batchelor (1927: 14, see below). If the reader wishes to check the quotation or read the context in which it was reported, the complete citation can be found by looking up the Ainu in the HRAF bibliography. I fol­lowed this procedure to shorten the length of the bibliography. In one or two instances, where an ethnographic account not pro­cessed by HRAF was used, the source in question is cited with the other references in this paper.

Simmons' measure of female status: "Subjection or inferiority of women," was coded by him on a four-point scale: (1) "absence or non-appearance of the trait, when definitely indicated in the sources;" (2) "incipient presence, slight elaboration, or cultural unimportance of the trait;" (3) "presence without dominance, moderate elaboration, or intermediate importance of the trait;" (4) "dominance, marked elaboration, or strong social importance of the trait in the particular tribe under investigation" (see Simmons 1937: 497). For analysis, Simmons' scale was collapsed into two ranks: (1) his ranks 1 and 2, and (2) his ranks 3 and 4.

My measure of female status is made from the ethnographers' quotes listed below for each society. A code of 1 means that fe­male status was interpreted to be equal to male status or only slightly below; a code of 2 means that female status was consid­ered to be markedly below male status.

Cultural evolution is measured by the maximum settlement population in a society. The name and population of the largest settlement in the society are provided. For analysis, the popula­tions were ranked into a nine-point scale (see Tables 1 and 2). The time periods for which the female status measures and the settlement size measures apply are also listed. Information is also provided if evidence was found in a source (1) of a field stay of one year or more, and/or (2) that the ethnographer could speak the native language, or if (3) the source was listed in the "Stan­dard Ethnographic Sample" (Naroll and Sipes 1973), which in-

Female Status and Cultural Evolution 199

eludes only authors who have met both the field stay and language familiarity requirements. The sex of the ethnographer can be as­certained by consulting the HRAF Source Bibliography. The five societies for which at least one of the observers was female are: Albanians, Kung Bushmen, Navaho, Omaha, and Samoa.

Ainu AB6 Simmons' code of female status: 2 Author's code of the female status: 1 Time: 1877-1924 "The women, where they had not been touched by outside in­fluences, were regarded and treated as the equals of men" (Bat­chelor 1927: 14). Settlement size: 1,669 Time: 1888 Village of Hidaka (Hitchcock 1891: 438-39) Length of field stay and knowledge of native language: forty­seven years, from 1877 to 1924 (Batchelor 1927: xii); spoke the language (Batchelor 1927: 59, 205, 279).

Albanians EGl Simmons: 2 Author: 1 Time: 1910 "Her status is, I believe, unique. She associates with men on equal terms and eats and smokes with them" (Durham 1928: 194). Settlement size: 35,000 Time: 1936 Town of Tirana (Redlich 1936: 87)

Andamanese AZ2 Simmons: 1 Author: 1 Time: 1869-1908 "Women may occupy a position of influence similar to that of the men" (~adcliffe-Brown 1922: 47). "One of the most striking features of their social relations is the marked equality and affection which subsists between husband and wife" (Man 1923: 41). Settlement size: 218 Time: 1901 Yere band (Temple 1903: 6) Field stay and language familarity: Radcliffe-Brown stayed over a year and spoke the native language (cited in Naroll and Sipes 1973: 116).

Araucanians SG 4 Simmons: 2 Author: 2 Time: 1952-54 "Compared to men, women have a generally inferior position in Mapuche society" (Faron 1968: 32).

200 Behavior Science Research 3, 1976

Settlement size: 50,000 Time: 1952-1954 Town ofTemuco (Faron 1961: 8) Field stay from 1952 to 1954 (Faron 1968: vii). He did not speak Araucanian and used an interpreter (Faron 1968: 7).

Ashanti FE 12 Simmons: 1 Author: 2 Time: 1945 "Though women in theory have the same economic opportunities and legal status as men, in practice their freedom is less" (Fortes 1949: 68). Settlement size: 70,705 Time: 1948 Town of Kumasi (Busia 1951: 125)

Aztecs NU 7 Simmons: 2 Author: 2 Time: ca. 1520 "Women had definite rights but they were inferior to those of men" Vaillant (1941: 112). Settlement size: 400,000 Time: ca. 1565 City of Tenochtitlan (Cook and Simpson 1948: 34F)

Bushmen FX 10 Simmons: 1 Author: 2 Time: 1952-61 "In Kung society, however, the magnitude of their [women's] contribution does not make women dominant. There are factors which bring the men forward in dominance and leadership" (Marshall 1959: 363). "The women play a dependent role in Kung society" (Marshall 1965: 255). Settlement size: 60 Time: 1951-61 "Most of the Nyae Nyae bands were composed of several families and ranged in size from 20 to 25 to 50 or 60 members" (Marshall 1965: 267). Field stay: participation in six expeditions, from 1951 to 1961. (Marshall 1958: 49).

Chippewa NG 6 Simmons: 2 Author: 2 Time: 1777-1804 "The women are considered as mere slaves to their husbands" (Camerson 1890: 257). "The moment she becomes a wife, she loses her liberty and is an obsequious slave to her husband who never loses sight of his

Female Status and Cultural Evolution 201

prerogative" (Long 1904: 174). Settlement size: 420 Time: 1736 Auwause band (Hickerson 1962: 33) Field stay and language familiarity: eighteen years, from 1786 to 1804 (Cameron 1890); and four years, from 1777 to 1781 (Long 1904). Long spoke the language well (Dunning 1959: 80).

Chukchee RY 2 Simmons: 2 Author: 2 Time: 1890-1901 "Women, whose social position is much inferior to that of men ... " (Bogoras 1904-09: 537). Settlement size: 299 Time: 1890-1901 Ne' ekan camp (Bogoras 1904-09: 27-32) Field stay: eleven years, from 1890 to 1901. He spoke the language (Bogoras 1904-09, cited in Naroll and Sipes 1973: 116).

Crow NQ 10 Simmons: no data Author: 2 Time: 1833-56 "It has ever been the custom of these wandering people to regard females in an inferior light in every way" (Denig 1961: 195). Settlement size: 6,400 Time: 1833 "The entire Crow nation banded together for the buffalo hunt. In 1833 three were 800 lodges with an average of 8 persons per lodge" (Denig 1961: 169).

Fang FH 9 Simmons: 2 Author: 2 Time: 1935-54 "The woman has no authority in the household; she is hardly more respected than a domestic animal'' (Trezenem 1936: 28). "Very often the role of women is secondary ... in short they only carry out men's orders" (Alexandre and Binet 1958: 105). Settlement size: 4,125 Time: 1946-51 Town of Effak (Baladier 1955)

Hottentots FX 13 Simmons: 1 Author: 1 Time: 1930 "The status of the wife in Hottentot society is far from b_eing that of an inferior. Although as a rule she plays a subordinate role in matters pertaining to tribal life . . . yet her position in the house­hold is supreme" (Schapera 1930: 251). Settlement size: 400 Time: 1933

202 Behavior Science Research 3, 1976

Chainouquas band (Dapper 1933: 27) Language familiarity: Schapera was an expert in Bushmen lan­guages (Schapera 1929: 144-62).

Iban OC 6 Simmons: 1 Author: 1 Time: 1949-58 "Between husband and wife there is no clearly structured pattern of superordination-subordination" (Freeman 1958: 33). This pat­tern appears to also have existed in earlier times as well, as per Roth (1891: 127) and Gomes (19i1: 86). Settlement size: 169 Time: 1958 Village of Pruan (Bedford 1959: 204) Field stay and language familiarity: Freemen stayed over a year and spoke the language (cited by Naroll and Sipes 1973: 117).

IncaSE 13 Simmons: no data Author: 2 Time: 1610 "And among those people the women were so subjected and so pressed into serving their husbands and to following their will ... etc. There was no manner in which she could escape from her husband's bondage except through death" (Coho 1893: 180). Settlement size: 200,000 Time: ca. 1530 City of Cuzco (Murdock 1934: 423)

Iroquois NM 9 Simmons: 1 Author: 2 Time: 1844-50 "The Indian regarded woman as the inferior, the dependent, the servant of man, and from nature and habit, she actually considered herself to be so" (Morgan 1901: 1,315), Settlement size: 3,000 Time: ca. 1687 "It is not improbable that the largest villages of the Iroquois contained 3,000 inhabitants" (Morgan 1901: 1,308; 2,229). Field stay and language familiarity: six years, from 1844 to 1850 (Morgan 1901: 1; xi). The ethnographer used an interpreter (Mor­gan 1901: 2,154).

]ivaro SD 9 Simmons: 1 Author: 1 Time: 1916-29 "The social position of the Jibaro woman is by no means that of an oppressed slave. She has much to say even in matters which

Female Status and Cultural Evolution 203

do not directly concern her own domain within the household" {Karsten 1935: 223). "The truth is that the married J ibaro woman is not only completely independent within her own sphere of activity, but exercises a remarkable social influence and authority even in matters which mainly concern her husband" (Karsten 1935: 253-54). " ... hence the extraordinary importance and the great social in­fluence of the Jibaro woman" {Karsten 1935: 256). Settlement size: 195 Time: 1946-48 "Geographical group" of Yaupi {Daniels son 1949: 88) Field stay and language familiarity: four years, from 1916 to 1918 and 1928 to 1929 {Karsten 1935: 540). The ethnographer spoke the language (Karsten 1935: 540).

Lapps EP 4 Simmons: 1 Author: 1 Time: ca. 1935-55 "The position of the wife has always been relatively good, al­though women have been excluded from certain hunting rituals and sacrificial ceremonies" (Indian University 1955: 44). "The married woman seems to have had a more independent status among the Lapps than among the Scandinavians" (Collinder 1949: 131). Settlement size: 196 Time: 1943-45 Koukama {Manker 1953: 21) Field stay: three years {Collinder 1949: v).

Maori OZ 4 Simmons: 2 Author: 2 Time: 1870-1920 "The female sex is given an inferior position" {Best 1924: 93). "The male sex was generally considered superior to the female" {Best 1924: 361). "The position of inferiority assigned to the sex [female] was as­suredly an unpleasing feature ... " (Best 1924: 404). Settlement: 134 Time: 1938 Kahukura (Hawthorn 1944: 25) Field stay and language familiarity: fifty' years in New Zealand {Best 1924). The ethnographer translated native songs (Best 1924: 136, 139-40).

Navaho NT 13 Simmons: 1 Author: 1 Time: 1941-49

204 Behavior Science Research 3, 1976

"Despite the absence of the symbols which whites associate with high status of women, however, there can be no doubt that the position of women among the people is very good" (Kluckhohn and Leighton 1946: 56). "There can be no doubt that the status of women among the people is very good" (Henry 1947: 116). " ... the position of women was high in respect to decisions and informal inter-relations" (Landgraf 1950: 217). Settlement size: 600 Time: 1947 Ramah Navaho, in an area east of Zuni reservation (Adair and Vogt 1949: 551) Language familiarity: Kluckhohn was moderately fluent (Leighton and Leighton 1949: V).

Omaha NO 12 Simmons: 1 Author: 1 Time: 1871-1905 "Men and women were socially on a moral equality" (Fletcher and LaFlesche 1906: 242). "Politeness is. shown by men towards women"(Dorsey 1881: 270). Settlement size: 1,179 Time: 1884 The tribe as a whole is considered, because they lived together during the summer hunt (Dorsey 1884: 284; Fletcher and La Flesche 1911: 33). Field stay and language familiarity: Fletcher spent twenty-nine years, from 1875-1905, and La Flesche was a native Omaha (Flet­cher and La Flesche 1911: 29). Dorsey wrote a book on the Omaha language (1890). Fletcher used the language, and La Flesche was a native speaker (Fletcher and La Flesche 1911: 605-07).

Pomo NS 18 Simmons: 1 Author: 1 Time: 1892-1918 "The status of Porno women was rather high. . . . The impression is one of definitely greater social equality of the sexes than among the northwestern tribes" (Kroeber 1953: 256). "In other words, there is patent in Porno ceremonies a rather thorough going democracy regarding the positions of the sexes" (Barrett 1917: 441). Settlement size: 325 Time: 1924 Cigom band (Loeb 1926: 365) Field stay and language fall)iliarity: Barrett stayed over a year and spoke the language (cited in Naroll and Sipes 1973: 127).

Female Status and Cultural Evolution 205

Samoans OU 8 Simmons: 1 Author: 1 Time: 1925, 1937 "The public role of a woman is entirely confined to wire-pulling and the private manipulation of the men-folk within her sphere of influence. That sphere is often very wide, as a woman is able to use her manual skill as well as her knowledge of intrigue in obtaining recognition within the household" (Mead 1928: 483). "The status of women in Samoa today, while in outward ceremonial matters remaining much as before, has changed subtly in the di­rection of giving them greater independence and power" (Keesing ·1937: 8). Settlement size: 956 Time: 1926 The largest of 236 villages from the four Islands of Samoa con­tained 956 persons in 1926 (Keesing 1934: 41).

Semang AN 7 (See also Malaya AN 1.) Simmons: 1 Author: 1 Time: 1924 "Marriage is based on equality of rights between man and woman; both enjoy complete freedom and equal rights which are observed upon the contraction of a marriage, during the whole time it lasts, and also after separation" (Schebesta 1927: 279). Settlement size: 54 Time: 1924 Teladn band (Schebesta 1927: 104, 109, 170, 207) Field stay and language familiarity: Schebesta spent twenty-one months, from January 1924 to September 1925 (Evans 1937: 12). He was learning the native language, spoke in a contact language, and did not use an interpreter (Schebesta 1927: 92; 1954: 3).

Seri NU 31 Simmons: 1 Author: 1 Time: 1894 "The most noticeable social fact revealed about the Seri rancher is the prominence of the females, especially the elderwomen in the management of everyday affairs" (McGee 1898: 269). " ... the somewhat exceptional manifestation of property right in the females, the singularly strong sense of maternal relation, and the apparent prominence of females in shamanistic practices as well as in the tribal councils" (McGee 1898: 274). Settlement size: 215 Time: 1955 Unnamed group (Marroguin 1957: 339)

Shilluk F] 23 Simmons: 1

206 Behavior Science Research 3, 1976

Author: 1 Time: 1910 "The position of the woman is no doubt a higher one than with most Mohammedan people of the Sudan. She is generally well treated and is shown remarkable respect" (Westermann 1912: 111). Settlement size: 720 Village of Atwadoi (Westermann 1912: xxii)

Tasmanians 01 19 Simmons: 2 Author: 2 Time: ca. 1890 " ... they did not treat their women well. La Billaridiere states, 'It gave us great pain to see these poor women condemned to such severe toil . . . ' and Peron, describing a meeting with 20 female aborigines, says, 'They were nearly all covered with scars, the miserable results of the bad treatment of their brutal husbands' " (Roth 1890: 256). Roth (1890: 126) quotes La Billardiere as saying, "The women showed the greatest subordination to their husbands ... the wom­en, who in 9 cases out of 10, were no better than slaves." Settlement size: 48 Time: ca. 1890 Unnamed band (Roth 1890: 28)

Toda AW 60 Simmons: 2 Author: 1 Time: 1870-71 "They [Toda women] are treated with respect, and are permitted a remarkable amount of freedom" (Marshall 1873: 43). " ... a remarkable symbolic acknowledgment of the exceptionally important position which the Toda women occupy" (Breeks 1873: 25). Settlement size: 507 Time: 1871 Village of Todanad (Breeks 1873: 5)

Trobriands OL 6 Simmons: 1 Author: 1 Time: 1914-20 "The position of women among them is by no means characterized by oppression or social insignificance" (Malinowski 1922: 380). "In all concrete manifestation of rank, whether traditional or social, the 2 sexes are equal" (Malinowski 1929: 30). . Settlement size: 325 Time: 1950 Village of Omarakana (Powell 1960: 120-21) Field stay and language familiarity: four years, from 1914-20 (Ma­linowski 1922: xvii). The ethnographer spoke the language (Ma­linowski 1922: xvi).

Female Status and Cultural Evolution 207

Tuareg MS 25 Simmons: 1 Author: 1 Time: 1905-54 "The woman holds a privileged position in Taureg society" (Blan-quernon 1955: 145). Time: 1937-54 "In life, the woman enjoys a freedom of conduct which is not customarily allowed in Moslem countries'' (Benhazera 1908: 9).

Time: 1905 "Although the woman ascribes the caste of her children and as­sures her political rights, her social position is not predominant in proportion to that of the man, who is the head of the family" (Lhote 1944: 5). Time: 1929-40 Settlement size: 686 Time: 1938 Agricultural center of Tamanrasset (Lhote 1944: 105)

Vedda AX 5 Simmons: 1 Author: 2 Time: 1906-18 "The females are kept in a state of rigid subjection. to the males" (Lewis 1919: 124). Settlement size: 308 Time: ca. 1863

. Bintenne band (Bailey 1963: 296)

Yahgan SH 6 Simmons: 1 Author: 1 Time: 1918-24 "The important position of the woman in the economy assures her position within the family equal to that of the man" (Gusinde 1937: 627). "Within the family the woman's position is almost exactly on a par with that of the man" (Gusinde 1937: 1089). Settlement size: 258 Time: 1924-25 Hoste Island band (Lothrop 1928: 196) Field stay and language familiarity: six years, from 1918 to 1924 (Gusinde 1937: 1460. The ethnographer spoke the language (Gu­sinde 1937: 1459).

REFERENCES*

Baker, Tanya, and Mary Bird 1959 "Urbanization and the position of women," Sociological Re­

view n.s. 7: 99-122. Barnouw, V.

1963 Culture and personality, Homewood, Ill., Dorsey Press.

* For complete ethnographic references, see HRAF 1976.

208 Behavior Science Research 3, 1976

Benedict, Ruth 1934 Patterns of culture, New York, Houghton Mifflin.

Benston, Margaret 1969 "The political economy of women's liberation," Monthly Re­

view 24: 13-27. Bertner, Margaret B.

1974 Feminine status in West African societies: a problem in con­cept definition, paper presented at the Annual Meeting, Ameri­can Anthropological Association (Mexico City).

Blalock, Hubert M., Jr. 1972 Social statistics, 2d ed., New York, McGraw-Hill.

Boserup, Ester 1970 Woman's role in economic development, New York, St. Mar­

tin's Press. Bowden, Edgar

1969 "An index of sociocultural development applicable to pre­civilized societies," American Anthropologist 71: 454-61.

1972 "Standardization of an index of sociocultural .development for precivilized societies," American Anthropologist 74: 1122-32.

Carneiro, Robert L. 1970 "Scale analysis, evolutionary sequences, and the rating of

cultures," in Roul Naroll and Ronald Cohen, eds., A Handbook of Method in Culiural Anthropology, Garden City, N.Y., Natu­ral History Press: 834-71 (reissued 1973 by Columbia Univer­sity Press).

Carneiro, Robert L., and Stephen F. Tobias 1963 "The application of scale analysis to the study of cultural evolu­

tion," New York Academy of Sciences, Transactions 26: 196-207. Divale, William Tulio

1974 The causes of matrilocal residence: a cross-ethnohistorical survey, Ann Arbor, Mich., University Microfilms (No. 75-7742).

1975 "Temporal focus and random error in cross-cultural hypothesis tests," Behavior Science Research 10: 19-36.

Driver, Harold E. 1970 "Statistical studies of continuous geographical distributions,"

in Raoul Naroll and Ronald Cohen, eds., A Handbook of Method in Cultural Anthropology, Garden City, N.Y., History Press: 620-48 (reissued 1973 by Columbia University Press).

Evans-Pritchard, E. E. 1965 "The position of women in primitive societies and in our own,"

in his The Position of Women in Primitive Societies, and Other Essays in Social Anthropology, New York, Free Press: 37-58.

Fortune, Reo F. 1939 "Arapesh warfare," American Anthropologist 41: 22-41.

Freeman, Linton C. 1957 An empirical test of folk-urbanism, Ann Arbor, Mich., Univer­

sity Microfilms, Dissertatum Abstracts 17: 3112. Freeman, Linton C., and Robert F. Winch

1957 "Societal complexity: an empirical test of a topology of socie­ties," American journal of Sociology 62: 461-66.

Female Status and Cultural Evolution 209

Goldberg, Marilyn Power 1970 "The economic exploitation of women," Review of Radical

Political Economics 2, no. 1. 1972 Women in the Soviet economy. Review of Radical Political

Economics 4, no. 3. Harris, Marvin

1968 The rise of anthropological theory, New York, T.Y. Crowell. Hitchcock, Romyn

1891 "The Ainos of Yezo, Japan," U.S. National Museum, Annual Report (1889/1890): 429-502.

Hochschild, Arlie Russel 1973 "A review of sex role research," American journal of Sociology

78: 1011-29. Hogeland, Ronald W.

1971 "'The female appendage': feminine life-styles in America, 1820-1860," Civil War History 17: 101-14.

Human Relations Area Files (HRAF) 1976 HRAF source bibliography: cumulative, New Haven, Human

Relations Area Files (bibliography of the sources processed for the HRAF Files).

Knudsen, Dean D. 1969 "';[he declining status of women: popular myths and the failure

of functionalist thought," Social Forces 48: 183-93. Laswell, Thomas

1965 Class and stratum, Boston, Houghton Miffiin. Lefaucheux, Marie-Helene

1962 "The contribution of women to the economic and social de­velopment of African countries," International Labor Review 86: 15-30.

Lewis, Oscar 1951 Life in a Mexican village, Urbana, University of Illinois Press.

Little, Kenneth 1973 African women in towns, London, Cambridge University Press.

Marsh, Robert M. 1967 Comparative sociology, New York, Harcourt, Brace and World.

Mead, Margaret 1938- "The mountain Arapesh," 5 vols., Anthropological Papers

49 of the American Museum of Natural History. Murdock, George Peter

1934 "The Incas of Peru," in his Our Primitive Contemporaries, New York, Macmillan: 403-50.

Murdock, George Peter, and Caterina Provost 1973 "Measurement of cultural complexity," Ethnology 12: 379-92.

Naroll, Frada, Raoul Naroll, and Forrest H. Howard 1961 "Position of women in childbirth: a study in data quality con­

trol," American journal of Obstetrics ar~d Gynecology 82: 943-54.

Naroll, Maud 1974 Attila's sister: women's occupations in preindustrial society,

unpublished manuscript, State University of New York at Buf­falo.

210 Behavior Science Research 3, 1976

Naroll, Raoul 1956 "A preliminary index of social development," American An­

thropologist 58: 687-715. 1962 Data quality control-a new research technique, New York,

Free Press of Glen.coe. 1970 "Data quality control in cross-cultural surveys," Raoul Naroll

and Ronald Cohen, eds., in A Handbook of Method in Cultural Anthropology, Garden City, N.Y., Natural History Press: 927-45 (reissued 1973 by Columbia University Press).

Naroll, Raoul, and William Tulio Divale 1976 "Natural selection in cultural evolution: warfare versus peace­

ful diffusion," American Ethnologist 3: 97-129. Naroll, Raoul, and Richard G. Sipes

1973 "A standard ethnographic sample: second edition," Current Anthropology 14: 111-40.

Nie, Norman H., C. Hadlai Hull, Jean G. Jenkins, Karin Steinbrenner, and Dale H. Bent

1975 SPSS: statistical package for the social sciences, 2d ed., New York; McGraw-Hill.

Otterbein, Keith F. 1972 Comparative cultural analysis, New York, Holt, Rinehart and

Winston. Pelto, Pertti

1970 Anthropological research: the structure of inquiry, New York, Harper and Row.

Redfield, Robert 1930 Tepoztlan, Chicago, University of Chicago Press.

Rohner, Ronald P., Billie R. DeWalt, and Robert C. Ness 1973 "Ethnographer bias in cross-cultural research: an empirical

study," Behavior Science Notes 8: 275-317. Sanday, Peggy R.

1973 "Toward a theory of the status of women," American Anthro­pologist 75: 1682-1700.

Schaefer, James M. 1969 "A comparison of three measures of social complexity," Ameri­

can Anthropologist 71: 706-08. 1973 A hologeistic study of family structure and sentiment, super­

natural beliefs, and drunkenness, doctoral dissertation, State University of New York at Buffalo (University Microfilms, No. 73-29, 131).

Schapera, Isaa9 1929 "Bushmen languages," Encyclopaedia Britannica, 14th ed.,

1: 144-62. New York, Encyclopaedia Britannica. Schlegel, Alice

1972 Male dominance arid female autonomy: domestic authority in matrilineal societies, New Haven, HRAF Press.

Schuller, Mary Craig, and Elizabeth H. Wheeler 1959 The role of women in Africa, New York, African-American

Institute (Report of conference convened by the Women's Africa Committee at International House, N.Y.C., November 20-21, 1959).

Female Status and Cultural Evolution 211

Simmons, Leo W. 1937 Statistical correlations in the science of society," in George

Peter Murdock, ed., Studies in the Science of Society, New Haven, Yale University Press: 495-517.

Sumner, William G., and A. G. Keller 1927 The science of society, 4 vols., New Haven, Yale University

Press. Tatje, Terrence A., and Raoul Naroll

1970 "Two measures of societal complexity: an empirical cross­cultural comparison," in Raoul Naroll and Ronald Cohen, eds., A Handbook of Method in Cultural Anthropology, Garden City, N.Y., Natural History Press: 766-833 (reissued 1973 by Columbia University Press).

Trey, J. E. 1972 "Women in the war economy-World War II," Review of Radi­

cal Political Economics 4, no. 3. Westermarck, Edward

1905 "The position of women in early civilization," American jour­nal of Sociology 10: 408-21.

Witkowski, Stanley R. 1975 Ethnographic fieldwork: optimal versus nonoptimal condi­

tions, paper read at the Annual Meeting of the Society for Cross­Cultural Research (Chicago).