Do Mothers or Fathers Influence Children's Political Views More?

Meggie Shurcliff

TABLE OF CONTENTS:

I. Abstract - page 1

II. The Study - page 2

Research Question - page 2

Sampling - page 2

Data - page 3

Significance Tests - page 3

III. Discussion - page 9

Weaknesses - page 9

Extrapolation - page 10

IV. Conclusion - page 10

I. Abstract

"Do Mothers or Fathers Influence Children's Political Views More?" is a study attempting to determine the degree to which mothers and fathers influence their children's political views. The research question being posed was: do mothers or fathers influence their children's political views more? The researcher performed two chi-square tests of independence: one to test whether or not there is a relationship between mother's political views and children's political views, and the other to test whether or not there is a relationship between father's political views and children's political views. The hypothesized point was that fathers would have a stronger influence on their children's political views because traditionally (and stereotypically), the father is the more dominant and influential figure in the household.

The population considered in this study was the entire sophomore class at Buckingham Browne and Nichols School (BB&N), a small, private college preparatory school in Cambridge, Massachusetts. The sample consisted of all the students that returned the survey distributed to them asking them their political views, and those of their parents. The only population broader than that of the sophomore class at BB&N to which the sample results can be extrapolated is the population of all high school students at BB&N.

The results of both Chi-Square tests of independence performed indicated that there is an extremely strong relationship both between mother's and children's political views and between father's and children's political views. The sample size was 57, and the p-values of both tests were approximately zero; however, that of the test between father's political views and children's political views was very slightly smaller, indicating that in terms of politics, both parents exert a great influence on their children, but that of the father is more dominant.

Since the sampling method employed in this study was a voluntary response survey, there are many possible sources of bias that may have affected the results. This study could be improved with a larger budget and a greater time frame; the samples should be obtained through a simple random sample, and the parents of students should be asked about their political views directly, as opposed to having the students ask them and then recording their responses.

II. Study

1. RESEARCH QUESTION: Do fathers or mothers influence their children's political views more?

2. SAMPLING: The sample consisted of the students who responded to surveys distributed to the entire sophomore class at Buckingham, Browne and Nichols school (BB&N), a small, private college preparatory school in Cambridge, Massachusetts. The surveys asked the students the following

1. Politically, which of the following do you identify with?

2. Which does your mother identify with?

3. Which does your father identify with?

For each question, students were given the following options and instructed to circle the one that best answers the question:

Conservative Moderate Liberal Other / none.

The surveys were placed in the mailboxes of every member of the sophomore class at BB&N. The students were instructed to bring the surveys home to ask their parents how they identified themselves politically to avoid students guessing haphazardly and making errors, thus adding bias and stealing validity from the study. Then, the students returned the completed surveys to the researcher's mailbox, and the results were counted. Students were encouraged numerous times through e-mail, announcements, and other means to respond to the survey; this constant reminding was an integral part of the procedure which served to reduce bias caused by nonresponse.

3. DATA

	Mother's Political Views
Children's Political Views	Liberal	Moderate	Conservative	Other/ None
Liberal	18	7	1	0
Moderate	8	5	5	0
Conservative	3	5	3	0
Other	0	1	0	1

	Father's Political Views
Children's Political Views	Liberal	Moderate	Conservative	Other/ None
Liberal	21	3	2	0
Moderate	4	9	5	0
Conservative	1	4	6	0
Other	0	1	0	1

4. SIGNIFICANCE TEST

A. Mother's Political Vies vs. Child's Political Views

1. Name of Test:

Chi-square test for independence of mother's political views and student's political views

2. Hypotheses:

Null Hypothesis: There is no association between mother's political views and children's political views.

Alternative Hypothesis: There is a relationship between mother's political views and children's political views.

3. Conditions:

1) Simple Random Sample required only if extrapolating to a larger population. The sampling method employed in this study was a voluntary response survey, so an SRS was not implemented. This shows us that it would be inappropriate to extrapolate the results of this test to a larger population (for example, all private high school students and their parents in the U.S.).

2) No more than 20% of expected counts are less than 5. This is not fulfilled either: in fact, 75% of expected counts are less than 5.

3) All expected counts are 1 or greater. This is not fulfilled; 6 out of the 16 cells had expected counts less than 1.

Although none of the conditions are fulfilled, we will continue with the significance test anyway; however, this must be kept in mind when considering the results of this test.

4. Table of Observed/Expected/Components of Chi-Square values:

Student's Political Views	Mother's Political Views
Student's Political Views	Liberal	Moderate	Conservative	Other/ None	Total
Liberal	18/14.59 / .796	7/6.84 / .0037	1/4.11 / 2.3533	0/.456 / .456	26
Moderate	11/10.1 / .0801	2/4.73 / 1.575	5/2.84 / 1.643	0/.316 / .316	18
Conservative	3/6.18 / 1.636	5/2.89 / 1.54	3/1.74 / .91241	0/.193 / .193	11
Other	0/1.12 / 1.12	1/.526 / .4271	0/.316 / 3.16	1/.0351 / 26.5	2
Total	32	15	9	1	57

Note: The numbers in the table in plain font (the first number out of the three in each cell, separated by a slash) indicate the observed count of children who fall into each category in each cell. The numbers in italics indicate the expected count, found by multiplying the total of each row by the total of each column, and then dividing that value by the grand total, 57. The expected count is the value that would be expected in each cell of the table if there were no relationship between mother's political views and children's political views. In other words, if children's political views were completely independent from mother's political views, we would expect there to be about 15 students who were liberal whose mothers were liberal as well. The numbers that are in bold indicate the component of chi-square for each cell; that is, the observed value (the number in plain font) minus the expected value (number in italics) squared, and then that sum divided by the expected value. This value, called the component of chi-square, measures how far the observed counts fall from the expected counts.

5. Chi-square value, p-value, and degrees of freedom:

Chi-square = 39.89; p-value = .00000796; degrees of freedom = 9 (found by number of columns minus one times number of rows minus one) The p-value states that the probability of obtaining such extreme results if there is in fact no relationship between mother's political views and student's political views is approximately zero.

6. Graph:

The shaded region on the graph corresponds to the probability of obtaining such extreme results if in fact the null hypothesis is true and there is no relationship between mother's political views and children's political views. As is evident in the graph, the shaded area is so small it cannot even be seen.

7. Conclusion:

This sample provides extremely strong evidence (chi-square = 39.89, degrees of freedom = 9, p-value = approximately 0) to reject the null hypothesis which states that there is no relationship between the variables mother's political views and student's political views. This suggests that knowing a mother's political view can tell us something about her teenage child's political views. However, since the conditions for this test of significance were not fulfilled (see conditions section), these results may not be entirely accurate.

8. Follow-up Analysis:

The largest component of chi-square (~26.5) between the observed and expected counts occurred in row 4, column 4; it was the cell for which both the student and the mother identified themselves with the "other/none" category. The expected value of students who identify with the "other/none" category who also have mothers who identify with the same category was close to zero, while in fact there was one student in this category. The second largest component of chi-square (~2.35) occurred in the cell of liberal students with conservative mothers. Only one student fell into this category, although the expected count was much greater, at about four students.

B. Father's Political Views vs. Student's Political Views

1. Name of Test:

Chi-square test of independence of father's political views and student's political views.

2. Hypotheses:

Null Hypothesis: There is no relationship between the variables father's political views and children's political views.

Alternative Hypothesis: There is a relationship between the variables father's political views and children's political views.

3. Conditions:

2) No more than 20% of expected counts are less than 5. This is not fulfilled either: in fact, 62.5% of expected counts are less than 5.

3) All expected counts are 1 or greater. This is not fulfilled; 7 out of the 16 cells had expected counts less than 1.

Although none of the conditions are fulfilled, we will continue with the significance test anyway; however, this must be kept in mind when considering the results of this test.

4. Table of Observed/Expected/Component of Chi-Square values:

Children's Political Views	Father's Political Views
Children's Political Views	Liberal	Moderate	Conservative	Other/ None	Total
Liberal	21/11.85 /7.04	3/7.75/ 2.915	2/5.929/2.604	0/.456/.45614	26
Moderate	4/8.21/2.1592	9/5.368/2.456	5/4.105/.19501	0/.3157/.3157	18
Conservative	1/5.017/3.2168	4/3.28/.1577	6/2.508/4.8584	0/.1929/.1929	11
Other	0/.9123/.00843	1/.596/.27296	0/.4561/.45614	1/.0351/26.53	2
Total	26	17	13	1	57

Note: For explication of table (what the numbers in plain, italics, and bold font mean) can be found under the table of Mother's Political Views vs. Student's Political Views on the previous test of significance.

5. Chi-Square value, p-value, and degrees of freedom:

Chi-Square = 54.74922; p-value = .00000001359; degrees of freedom = 9. The p-value indicates that there is approximately zero chance of obtaining such extreme data values if father's political views and student's political views were independent, and no relationship existed between them.

6. Graph:

The shaded area of this graph indicates the probability of obtaining such extreme results if the null hypothesis was true and there is no relationship between father's political views and children's political views. The shaded area is so small it is not even shown on the graph, which means there is approximately zero chance of obtaining such results if there was no relationship between the two variables.

7. Conclusion:

This sample provides very strong evidence (chi-square = 54.75, p-value = approximately 0, degrees of freedom = 9) to reject the null hypothesis that there is no relationship between the variables Father's Political Views and Children's Political Views. This suggests there is a very strong association between father's political views and student's political views; that is, knowing something about a father's political views can tell us something about his children's. However, because the conditions were not fulfilled (see conditions section), these results may not be entirely accurate.

8. Follow-up Analysis:

The largest component of chi-square (approximately 26.5) is located in the cell for which both the student and the student's father identified with the "Other/None" category politically. This means that this cell contains the greatest difference between the observed value (1) and the expected value (.0351). This great difference between observed count and expected count indicates that more students than would be expected who identify with the "Other/None" category also have fathers who identify with the "Other/None."

III. Discussion

1. WEAKNESSES:

The most prevalent weakness in this study was the sampling design. As described above in the Study section, the sampling method was a voluntary response survey. Each of these sampling methods (voluntary response and surveys) opens the door for different types of bias to confound the results.

The voluntary response aspect of this sampling method creates bias because, as often is the case with voluntary response sampling, only those with strong opinions will respond. For instance, it is possible that liberals feel much more strongly about politics than conservatives, so maybe only the data for liberal students was recorded. If this is the case, it is possible that students who consider themselves conservative, moderate, or other all have liberal parents, or parents that display a significantly different trend in their responses than the results that were obtained. Additionally, perhaps every single student who did not respond to the survey was moderate; there were very few people who identified themselves as moderate, which may mean they don't feel as strongly either way about politics. Their opinions, however, are no less valuable than those of liberals or conservatives, so this could lead to all sorts of inaccurate results. There were a total of 57 students who replied to the survey, which is approximately half of the students to which the survey was distributed. While this is a significant amount of students, the other half of the sophomore class at Buckingham Browne and Nichols School and their parents could have political views that show different trends from the data obtained, so it is extremely likely that this study contains a fair amount of the bias caused by voluntary response.

Another source of error in this study is the fact that it employed a survey to obtain data. It is possible (although not very likely) that every single student either lied on the survey or wasn't sure of their parents' political opinions and guessed incorrectly. Although students were repeatedly and persistently encouraged to bring the surveys home to ask their parents themselves, it is possible that students did not follow the instructions and guessed randomly, perhaps assuming their parents were of the same political opinions as themselves. This would cause the bias of making the data seem much more significant than it actually is, because it would appear that both parents influence their children's political views much more than they really do. Students may have also haphazardly circled random choices without even looking at them and considering with which they really did identify.

Another significant weakness of this study is the fact that none of the conditions for either significance test were fulfilled. This is mainly due to the "Other/None" category, and the severe lack of students and parents that identified with it. Because of this, many cell counts were 0 or 1, and expected counts as well were frequently less than 1. Since the conditions were not met at all for either significance test, the results of the study must be taken with a grain of salt.

Thus, considering the flaws inherent in the sampling design and the many significant sources of error in this study, it is necessary to bear this in mind when considering the results of this study.

2. EXTRAPOLATION:

Since there was not a simple random sample implemented in the sampling design, it is not safe to extrapolate these results to larger populations. The only population to which these results can be extrapolated with some degree of comfort is the population of all sophomores at Buckingham Browne and Nichols School, since that is the only grade included in the sampling. Beyond this (and with much less confidence) these results could be extrapolated to the population of all students at the Buckingham Browne and Nichols high school, but that is the largest population to which these results could be extrapolated with any security. Had a simple random sample been implemented, these results would have been able to be extrapolated to larger populations such as all private high school students in Massachusetts, or even New England.

IV. CONCLUSION

From the two significance tests performed, it is clear that there is definitely a relationship both between father's and child's political opinion, and between mother's and child's political opinion. Both p-values, each at approximately zero, indicate that the relationship is very strong; however, the p-value for the test of the relationship between father's political opinions and child's political opinion is slightly smaller than that of the relationship between mother's political opinion and child's political opinion. This indicates that there exists a stronger relationship between fathers' political views and children's political views than between mother's political views and children's political views. To answer the research question, the results yielded indicate that fathers influence their children's political views more than mothers do. There are a number of reasons that could account for this. Traditionally, in a stereotypical household the father is the more dominant parent, while the mother quietly does the household chores and cares for the children. Perhaps this stereotype still holds true in today's modern world; maybe fathers are more vocal about their political beliefs, causing their children to share them, and either mothers aren't as active in politics or they keep their opinions to themselves. It is also possible that children idolize their fathers more than they do their mothers in various different aspects, not just politically, and because of this respect and admiration, children wish to be more like their fathers. Another possible explanation is that more males than females responded to the survey distributed to them, and maybe boys are more influenced by their fathers and girls by their mothers.

However, the difference between the p-value for the relationship between mother's political views and children's political views was only .000007946. This shows us that the difference between the p-values of the two tests is negligible. There is an extremely strong relationship both between father's political views and children's political views and mother's political views and children's political views. The two relationships are approximately equally strong, but the one between father's political views and children's political views is very slightly stronger.

This study leads to several other extremely interesting ones. The most direct study to perform after this would be to determine whether or not fathers influence sons more than daughters, and whether or not mothers influence daughters more than sons. Another interesting study that comes out of this would be to observe in what other ways parents influence their children, both in terms of mothers influencing daughters, fathers influencing sons and vice versa; for example, in terms of musical preferences, feelings about religion, and various personality traits. It would also be interesting to perform this study again, but with a much bigger budget and a greater time span and do a formal simple random sample of not only sophomores, but of students of each of the four grades of high school, and not only at Buckingham, Browne and Nichols, but of private and public high schools around the country. Also, the parents would be asked directly instead of having their children ask them and record their response. More specific questions would be posed, too; the survey here would not only ask students and parents whether they identify themselves with "liberal," "moderate," "conservative," or "other/none," but both students and their parents would be asked a specific host of questions about all sorts of current events, ranging from opinions on the Bush Administration to wars in general to gay marriage to the upcoming election. Should this study be implemented, it would be much clearer who influences children more and in what ways. The study performed- "Do Mothers or Fathers Influence Children's Political Opinions More?" opens the door to all these studies and provides a fundamental backbone and simple attempt at answering this question.