Research Design In Social Research - Cross-Sectional Design

Front 1

What are the three distinctive features of cross-sectional design?

Back 1

The three distinctive features of cross-sectional designs are that; they have no time dimension, they rely on data that captures existing differences rather than change following an intervention and the groups classified by the design are based on existing differences rather than random allocation.

Front 2

What can and can't cross sectional designs measure

Back 2

Cross-sectional designs can only measure differences between groups, they cannot measure change.

Front 3

On what Basis must groups be constructed in cross-sectional design and what are the limitations of this?

Back 3

Groups must be constructed on the basis of existing differences within the sample, with the sample divided into groups according to the category of individual groups to which they belong. The limitation lies in the fact that, because we must rely on existing differences, there can be no random allocation to groups which means the groups may very well be different in other respects apart from the category of independent variable to which they belong.

Front 4

In cross sectional design, what determines the number of groups?

Back 4

The number of groups is determined by the number of categories of independent variable that you want to and are able to look at.

Front 5

What does a repeated cross-sectional design involve?

Back 5

This design involves collecting information at a number of different time points, but from a different but comparable sample (asking the same questions etc) at each point.

Front 6

How are changes tracked in cross-seectional designs?

Back 6

Patterns in each sample are compared with those of previous or subsequent samples.

Front 7

Repeated cross-sectional designs are a form of repeated measure design. What are the two main forms of repeated measure design and what are the differences between them?

Back 7

3. The two main forms of repeated measure designs are prospective and retrospective. Prospective designs are where we begin now and repeat the study at various different points of time in the future. Retrospective designs are where we draw on existing data to examine patterns of channge up to the preseent point

Front 8

What are the two major sources of threats to the internal validity of cross-sectional research design?

Back 8

Problems in establiishing cause without a time dimension and problems at the level of meaning

Front 9

Why is Identifying causaliity problematic in cross-sectional rsearch?

Back 9

2. Differences between groups may not be due to their difference with respect to their differing independent varriables according to which they have been classified but to another difference or combination of differences – this is the problem of confounding variables

Front 10

How is the problem of confounding variables dealt with in cross-seectional research?

Back 10

1. Differences between groups are removed statistically after the data has been collected in order to remove as many of the confoounding variables as possible and compare like with like. New groups are created out of the sample who, while differing according to the independent variable do not differ according to a possible confounding variable. What is done is that you test the hypothesis that differences between groups are due to another confouding variables. For instance, suppose that in a cross sectional study you examine the effect of the presence of children (the independent variable) on marital happiness. You find that couples with children are, on average, happier than couples without. However this might be because couples with children are, on average, older than couples without. You can test this hypothesis by looking, within your sample, at differences in marital happiness between childless couples and couples with children who are the same age. If it were that the reason couples with children are generally more maritally happy than couples without children is that couples with children are generally older then when comparing couples with children and couples without children who are of the same age then this difference should disapear. By doing this you control for a confounding variable. You find that, controlling for age, there remains a difference in marital hapiness and so you can conclude that this difference is not due to age.

Front 11

How does the number of variables you control for relate to the confidence you can have in any discovered relationship?

Back 11

The more variables you control for the more confidence you can have in any discovered relationship.

Front 12

What is the problem with controlling for variables?

Back 12

3. The problem with controlling for variables, in contrast to when members of groups are randomly allocated, is that you can only control for variables that you have thought of and for which you have information. - you can therefore never be sure that you have controlled for all relevant variables.

Front 13

How can cross-sectional research help eliminate variables as possible causes?

Back 13

Cross-sectional research can help eliminate possible variables a causes by showing that there is no correlation between the dependent variable and a possible confounding variable

Front 14

What is the most useful way to go about the causal analysis of cross-sectional data?

Back 14

1. The most useful way to go about the causal analysis of cross-sectional data is to draw flow charts or path diagrams that have been developed before cross-seectional data is used to draw causal conclusions. The task of the data analyst is then to evaluate how well the models fit the data.

Front 15

What is the difference betweeen apriori reasoning and ad hoc reasoning?

Back 15

2. A priori reasoning involves propposing, on the basis of theorietical considerations and previous research, that X will cause Y and that therefore the the two will be correlated. Ad-hoc reasoning is when a correllation is obseerved and a causal story is then made up to explain why the correlation exists.

Front 16

What is the problem with ad hoc reasoning?

Back 16

3. The problem with ad hoc reasoning is that we can always make up a story regardles of the correlation we find – it is one thinng to make up a story that is consistant with a correlation, it is quite another to anticipate, on the basis of reasoning and theory, that a correlation should exist and then find that it does.

Front 17

In what way is the issue of causal direction problematicin cross-sectional research.

Back 17

Even if a very strong correllation is found between X and Y we do not know if it is X that is causing Y or Y that is causing X.

Front 18

In what way can we deal with the problem of causal direction in cross-sectional design?

Back 18

The problem of causal direction in cross-sectional design can be dealt with by developing apriori causal models to test – if data is found that is consistent with a thesis then it can be concluded that another study with a design that can establish causal direction would be a good thing to do or we can ascertain which theory of causal direction is more likely.

Front 19

In what circumstances is the attribution of causal direction not probelmatic?

Back 19

3. The attribution of causal direction is not problematic when the independent variable is fixed (e.g. gender, race) is not subject to manipulation (e.g. age) or where time orderings are straight forward (e.g. parents education precedesthe type of job wee get).

Front 20

What are some of the problems with the kind of quantified analysis, in which variables are related to eachother, that is involved with cross-sectional research?

Back 20

1. It leaves out the complexities of the activity being examined, the actual processes of inteeraction in which human life is being lead. It also ignores the actor's interperetation of the events and how these impact upon their actions. As such, it is hard to understand why the independent and dependent variables are or are not related. Finally, by only dealing with parts of social life one looses sight of the context in which social activity takes place and wee are therefore likely to misunderstand the meaning of the behaviour.

Front 21

What are the ways in which the meaning and context of peoples actions can be taken to account in cross-sectional research?

Back 21

1. It leaves out the complexities of the activity being examined, the actual processes of inteeraction in which human life is being lead. It also ignores the actor's interperetation of the events and how these impact upon their actions. As such, it is hard to understand why the independent and dependent variables are or are not related. Finally, by only dealing with parts of social life one looses sight of the context in which social activity takes place and wee are therefore likely to misunderstand the meaning of the behaviour.