Child Development From Infancy to Adolescence. Laura E. Levine
be sure whether the changes you see are due to age or to differences in socioeconomic status.
Longitudinal design: A research design that follows one group of individuals over time and looks at the same or similar measures at each point of testing.
Longitudinal studies also lock researchers into using one set of measures even if better alternatives come along. If they switch measures during the study and find changes in the level of the outcomes they are tracking, the researchers cannot be sure whether the changes are attributable to the fact that the participants are older or the new measure is actually measuring something slightly different than the original measure. Despite these concerns, however, longitudinal research provides a very powerful way to look at developmental change.
A cross-sectional design studies multiple groups of participants who cover the age range of interest to the researcher. If you were interested in developmental changes between elementary and middle school, you could collect data from groups of participants who were 8 years old, 10 years old, and 12 years old. Then, by comparing the results from the groups, you could construct a picture of the changes that occur over that period of development. Because you collect all your data at the same time, you can do so in a relatively quick, cost-effective manner. Obviously participant dropout is not an issue because there is only a single data collection. Based on cross-sectional research, you will know that children of different ages show differences on the outcome you measured, but you won’t know why. We presume age changes are responsible, but we need to be careful when making these assumptions.
Cross-sectional design: A research design that uses multiple groups of participants who represent the age span of interest to the researcher.
One of the big challenges in cross-sectional research is that the different age groups in the study must be as similar as possible on any variable that might affect the study’s outcome. To make this point clear, here is an extreme example. Imagine you are interested in how self-esteem changes during the transition from elementary to middle school. To examine these changes, you use a group of 8-year-old students who attend a public elementary school in a disadvantaged neighborhood, a group of 10-year-old students from a private school with a religious affiliation, and a group of 12-year-old students from a suburban public school. Even if you found differences in self-esteem between the groups, could you correctly interpret them as age-related changes associated with school transitions? Clearly you could not. Because the groups came from such widely different school settings (and, therefore, it is likely that they differ from each other in a variety of other ways), you could not make any valid interpretation of these data. Differences between groups in actual cross-sectional research are much subtler than those in this example, but any difference that is not recognized and accounted for by the research can be a threat to the validity of the conclusions drawn from this type of research.
Another difficulty with cross-sectional research is that different age groups or cohorts have lived during different times in history. These differences create what is called a cohort effect. For example, people in their 70s are likely to be less skilled with computers than those in their 20s. This does not mean computer skills decline with age; rather it reflects the effects of the introduction of the home computer in the late 1970s, when those people now in their 70s were already more than 30 years old. The skill advantage of younger people reflects their use of the computer from a much younger age.
Cohort effect: Differences between age groups in a cross-sectional study that are attributable to the fact that the participants have had different life experiences.
Finally, sequential designs bring together elements of cross-sectional and longitudinal research design. This design uses several groups of people of different ages who begin their participation in the study at the same time, just as cross-sectional research does, and follows them over a period of time that can vary from a few months to many years, just as longitudinal research does. What makes this method different is the overlap of the groups’ ages. For example, if we were looking at children’s health over the age range from birth to 20 years, we could begin by assessing four different groups: infants, 5-year-olds, 10-year-olds, and 15-year-olds. After we repeat our assessment 5 years later (when the infants are 5 years old, the 5-year-olds are 10, and so on), we will have two different groups that were assessed at age 5, two assessed at 10, and two assessed at 15. Because we needed to follow the groups for only 5 years to cover our age span of interest, the risk of participants dropping out of the research study is lower than it would have been in a 20-year study and this reduces sample bias. We also reduce the time, money, and labor needed to conduct the study compared to a 20-year-long study of children’s health. Finally, if there were any cohort differences between the groups, those effects would become apparent when we compare the results of the overlapping groups who are the same age in different years.
Sequential design: A research design that uses multiple groups of participants and follows them over a period of time, with the beginning age of each group being the ending age of another group.
Even though sequential research offers advantages, the cohort groups still need to be as much alike as possible at the start of the study, and you still need to be able to track and reassess the groups at regular intervals, so sample attrition can still be a potential problem. Test your understanding of these three different developmental research designs by trying Active Learning: Developmental Research Designs.
Active Learning: Developmental Research Designs
Look at the chart to the right and answer the following questions:
1 If only Group A was tested in 2018 and 2023, what type of developmental research design would this be?
2 If only Testing Year 2018 was carried out, what type of developmental research design would this be?
3 If all the groups were tested in years 2018 and 2023, what type of developmental research design would this be?
Answers:
1 Longitudinal research design, because the same group of participants is followed and retested after 5 years.
2 Cross-sectional research design, because different age groups are tested and compared at the same time.
3 Sequential research design, because different age groups are compared and the groups also are followed and retested after 5 years.
Interpreting and Using the Results of a Study
After conducting research, we must still interpret the results. Our final understanding of what the study showed is greatly affected by the way we interpret the data, and two people may look at the same data and interpret them in different ways. In reaching conclusions, we must be careful not to generalize beyond the characteristics of the sample that participated in the research. We also need to remember that conclusions drawn from research are generalizations that apply to groups of individuals. As you learned in Chapter 1, there is a great deal of diversity among individuals within any group, so our conclusions will not apply to every single child but the fact that some children do not fit the general pattern does not invalidate the general conclusion.
Research results are tested using various statistical methods to determine whether the outcomes may have happened by chance. However, even if the results are statistically significant (that is, they did not happen by chance), you still might wonder whether they make any difference in the real world. As an example, consider research on the relationship between birth order and intelligence (Rohrer, Egloff, & Schmukle, 2015). IQ test scores of over 17,000 participants in three international studies found that the scores of firstborns in the sample were higher than the scores of second-borns at a statistically significant level of .0001 (meaning that there are only 1 chance in 10,000 that this difference is an accidental or chance finding). But before any firstborn readers of this text begin celebrating their intellectual superiority over their siblings, note that the difference in