Sample size affects the reliability of correlation coefficients in that larger sample sizes generally provide more accurate and stable estimates of the correlation. With a small sample size, the correlation coefficient can be more sensitive to random variation, which might lead to misleading results. In other words, a small sample may show a strong correlation by chance, while a larger sample can help confirm whether that correlation is genuine and reflects a true relationship between the variables. Thus, larger samples tend to produce more reliable and generalizable correlation coefficients.