What’s My Age Again?

4yearolds 
Kindergarteners 
Total 

1 
2.00 
1.36 
1.67 
2 
3.30 
2.00 
2.61 
4 
3.00 
2.40 
2.69 
6 
1.70 
1.73 
1.71 
8 
3.30 
2.45 
2.86 
10 
4.50 
2.82 
3.62 
12 
8.90 
4.05 
6.36 
16 
8.00 
5.28 
6.57 
19 
16.4 
8.91 
12.5 
21 
17.2 
9.55 
13.2 
28 
24.1 
16.9 
20.3 
34 
24.8 
15.2 
19.8 
43 
26.2 
16.8 
21.3 
50 
32.2 
25.8 
28.9 
63 
39.8 
27.6 
33.4 
Figure 1: Graph of Mean Absolute Error across Photo and Participant Age
Pattern of Estimates
A measure called the ratio error (RE) was used to determine the pattern of children’s estimates; it was calculated for each age by dividing the child’s estimate by the correct age. In this way, we were able to assess the ages children underestimated and overestimated, on average. If the RE was less than one, the estimate was less than the actual age, and if the RE was greater than one, the estimate was greater than the actual age. If the RE was exactly one, the child estimated correctly.
We performed a 2 x 15 (age of participant by age of photograph) repeated measures ANOVA on the ratio error data. We found that overall, children seemed to overestimate the younger ages and underestimate the older ages. They were most accurate at estimating the ages of sixyearolds, 21yearolds, and 28yearolds (see Table 2, Figure 2). The ANOVA results indicated a significant main effect of age of photograph in this case F (1,19)=7.16, p<.05). Using this measure, we have insufficient information to conclude that children’s accuracy increases with the age to be estimated; however, we can infer that the pattern of under and overestimation found in the descriptive statistics was significant. Neither the main effect of participant age nor the interaction between participant and photograph age was significant with F (1,19)=.04, p=.84 and F (1,19)=.25, p=.62 respectively.
Table 2: Relative Error across Age of Participant and Photograph
4yearolds 
Kindergarteners 
Total 

1 
2.80 
2.36 
2.57 
2 
2.45 
2.00 
2.21 
4 
1.25 
1.50 
1.38 
6 
1.02 
1.11 
1.06 
8 
0.84 
0.88 
0.86 
10 
0.95 
0.94 
0.94 
12 
1.13 
0.78 
0.95 
16 
0.59 
1.09 
0.85 
19 
1.17 
1.39 
1.29 
21 
1.03 
1.14 
1.09 
28 
1.13 
0.99 
1.06 
34 
0.84 
0.61 
0.72 
43 
0.79 
0.77 
0.78 
50 
0.36 
1.03 
0.71 
63 
0.66 
0.91 
0.79 
Figure 2: Graph of Relative Error across Age of Participant and Photograph
.
This experiment was designed to study the effect of age of participants on their ability to judge the ages of photographs of different ages. Participants in 3yearold, 4yearold, and kindergarten classrooms were asked to estimate the ages of 15 photographs of Caucasian males. We hypothesized that (a) older children would be more accurate at estimating the ages of photographs overall, (b) children would be more accurate at estimating the ages of photographs close in age to themselves, and (c) children would overestimate younger ages and underestimate older ages. These hypotheses all stem from the idea that children use themselves as an anchor—a representation of a known age—from which to adjust either up or down when estimating unknown ages.
Accuracy of Estimations
Overall, kindergarteners were significantly more accurate than 4yearolds. The data for the 3yearolds were not included due to their lack of ability to provide numerical estimations. The large majority of 3yearolds were only able to provide “older than me” or “younger than me” responses. Because of this, we concluded that both 4yearolds and kindergarteners do significantly better than 3yearolds at providing numerical estimations. In fact, this indicated that the ability to estimate numbers arises between the ages of three and four, after which estimation skills continue to be refined. The emergence of estimation may coincide with a sudden increase of mathematical ability or training (e.g., learning to count and name numbers greater than 10) between three and four years old. Siegler and Booth (2005) postulated that this significant improvement in estimation skills might be due to improvements in counting strategies. This was supported by the fact that in most 4yearold and kindergarten classrooms, children often practice counting by 1’s, 5’s, or 10’s every day. Children in 3yearold classrooms do not usually get such rigorous mathematical training.
Children were, on average, most accurate at estimating ages below and including eightyearolds. In particular, they were most accurate at estimating the ages of sixyearolds and oneyearolds. Accuracy got significantly worse as the age of the photograph increased. This confirmed the hypothesis that children are better at estimating the ages of other children. This result is supported by the finding of Britton and Britton (1969b) that it was more difficult for children to discriminate between the ages of adults because, to children, adults are lumped into one big category.
Additionally, the distribution of errors resembled a logarithmic distribution. This was consistent with the notion that children represent the number line logarithmically until approximately fourth grade (Siegler & Opfer 2003). In other words, children view distances between larger numbers as larger than distances between smaller numbers and so it would make sense that they also view the distances between larger ages as larger than the distance between smaller ages. In fact, Petito (1990) postulated that until third grade, children viewed numbers as ordinal rather than rational and so had no concept of the magnitude of the distance between two numbers. This implied that children would only be able to estimate ages of which they had prior knowledge and would not be able to accurately extrapolate up or down. In other words, children were accurate at estimating the ages for which they developed a mental representation but were unable to use those representations to estimate other ages because they were unable to determine the magnitude of the distances between ages.
Pattern of Estimates
The pattern of children’s estimates was assessed using the ratio measure. We found that the age of the photograph had a significant effect on children’s estimates, meaning that there were some ages that children consistently overestimated and some that they consistently underestimated. Overall, they overestimated the ages until about age six, underestimated between six and 16, overestimated between 16 and 28, and underestimated the ages after about age 28. Although we expected children to be accurate at around their own age, we did not expect the second rise in accuracy. It is possible that this was due to the use of a second reference point: their parents’ ages, in addition to their own age. Their estimates might be explained by breaking the ages into categories: younger than me, around the same age as me, older than me, younger than my parents, around the same age as my parents, older than my parents. When broken up in this way, children overestimated both categories in which the photographs were younger than someone. They were most accurate at guessing the ages of people that were around the same age as them or their parents, and they underestimated the ages that were either older than them, or older than their parents. This provided evidence that children might actually have two frames of reference, or anchors, when judging age. It is possible that children estimate age more accurately as they get older not only because of improved numerical aptitude but also due to the increased number of reference points they have. This may be because of new siblings, new older friends, or just people they already know (themselves included) getting older. These new points of reference may be additive, which allows them to estimate successfully the ages they already knew as well as the new age references that they acquire.
Although the results resembled the results we expected, this experiment could have benefited from more participants. There were some photograph age results that were heavily skewed by outlier data (e.g., a child estimating 16 years for the oneyearold). Using more participants would have more effectively balanced out the outliers. More participants would also allow for explorations into participant gender effects. This experiment did not take gender effects into account because, at the most, there would have been five girls and five boys in each group. With that few participants per condition, there would not have been enough power to get significant results.
The other gender effect we did not take into account was the gender of the photograph. Another experiment could be done to test the accuracy of children’s estimation of female ages. This effect was not tested in our experiment because adding another gender would mean twice as many photographs to show each child. We decided to sacrifice generalizability in order to be able to show each child a wider range of ages. Other experiments could either only use female photographs, or narrow down the number of photographs to a smaller set of essential ages.
Narrowing the set of ages may be helpful in and of itself because there were significant testing effects using 15 photographs. The children seemed to get tired or bored by around the 10 th picture. Although randomizing the order of the photographs shown minimized the effect this had on our data, better data would still be gained by either lowering the number of photographs or making the experiment more interesting and engaging in some way.
In summary, the results of this experiment supported all three of our main hypotheses. The ability to make estimations emerged between the ages of three and four and, in addition, older children were more accurate at estimating than younger children. Children also were most accurate at estimating the ages of photographs their own age and younger. In addition to supporting the hypothesis that when children used themselves as reference points, they underestimated people who were younger and overestimated people who were older, we also found that children used their parents’ ages as a reference point and that the same pattern applied. As children become older, they mature both in their ability to estimate and their ability to discriminate between different ages. It is possible this is due to the increased number of reference points they accumulate.
Britton, J. O., & Britton, J. H. (1969a). Discrimination and perception of age and aging by elementary school children. Proceedings of the 77 th Annual Convention of the American Psychological Association, 4, 715716.
Britton, J. O., & Britton, J. H. (1969b). Discrimination of age by preschool children. Journal of Gerontology, 24, 457460.
Edwards, C. (1984). The age group labels and categories of preschool children. Child Development, 55, 440452.
FGNET Aging Database . (n.d.). Retrieved February 25, 2008, from http://www.fgnet.rsunit.com/.
George, P., Hole, G., & Scaife, M. (2000). Factors influencing young children's ability to discriminate unfamiliar faces by age. International Journal of Behavioral Development, 24(4), 480491.
Jacobs, J., & Potenza, M. (1991). The use of judgment heuristics to make social and object decisions: A developmental perspective. Child Development, 62, 166178.
Jones, G., & Smith, P. (1983). The eyes have it: Young children’s discrimination of age in masked and unmasked facial photographs. Journal of Experimental Child Psychology, 38(2), 328337.
Kogan, N., Stephens, J., & Shelton, F. (1961). Age differences: A developmental study of discriminability and affective response. Journal of Abnormal and Social Psychology, 62(2), 221230.
Kratochwill, T., & Goldman, J. (1973). Developmental changes in children’s judgments of age. Developmental Psychology, 9(3), 358362.
Looft, W. (1971). Children’s judgments of age. Child Development, 42(4), 12821284.
Montpare, J., & McArthur, L. (1986). The influence of facial characteristics on children’s age perceptions. Journal of Experimental Child Psychology, 42(3), 303314.
Petitto, A. L. (1990). Development of numberline and measurement concepts. Cognition and Instruction, 7, 5578.
Siegler, R., & Opfer, J. (2003). The development of numerical estimation: evidence for multiple representations of numerical quantity. Psychological Science, 14(3), 237243.
Siegler, R. S., & Booth, J. L. (2005). Development of numerical estimation: A review. In J. I. D. Campbell (Ed.), Handbook of mathematical cognition (pp. 197212). New York: Psychology Press.
Smith, D. (1999). Use of the anchoring and adjustment heuristic by children. Current Psychology: Developmental  Learning  Personality  Social, 18(3), 294300.
Taylor, E., Steele, C., & Roberto, K. (1982). Preschool children’s discrimination of age. Perception and Motor Skills, 54(2), 539542.
