Why does my weighted average change depending on how I combine categories?

I’m getting myself tangled up with weighted averages and I can’t figure out why my overall number keeps changing depending on the route I take. In my class, quizzes are worth a chunk, projects are worth a bigger chunk, and exams are the rest. There are lots of little quizzes, a couple big projects, and a couple exams. If I take the average quiz score, the average project score, and the average exam score, then apply the listed weights and add them up, I get one overall grade. But if I think of every single item individually and try to give each quiz an equal slice of the quiz bucket, each project an equal slice of the project bucket, etc., I end up with a slightly different overall. I’m probably doing “percents of percents” in the wrong order.

It kind of feels like making a smoothie: should I blend each fruit type first and then mix those blends in the right proportions, or just toss every piece of fruit in at once and blend everything together? I keep ending up with two slightly different “flavors.”

What’s throwing me off is that each category has a different number of items and different point totals (like quizzes might be out of 10 each while projects are out of 100), and sometimes a lowest quiz gets dropped. Does averaging the category first and then weighting it actually match the idea of a weighted average, or should I be weighting every individual item according to its share of the total points? And when a lowest score is dropped, does that secretly change the category’s effective weight?

I feel like I’m close, but I keep second-guessing whether I’m double-counting something or watering something down. Could someone explain which method is the “right” one here and why my two approaches aren’t giving me the same number?

Any help appreciated!

3 Responses

  1. You’re getting two answers because you’re using two different grading models. In a category-weighted model (e.g., quizzes 20%, projects 30%, exams 50%), you should first compute each category’s average according to its internal rules (equal-per-item or points-proportional within that category, and after any drops), then multiply those category averages by the stated category weights and add. In a points-based model (“your grade is total points earned divided by total points possible”), you ignore categories and weight each item by its points; that’s equivalent to assigning each item weight equal to its points share of the course total. These two methods agree only if, within each category c, you spread the category weight W_c across items in proportion to their intended within-category weights (for example, by points), so each item i gets weight W_c × points_i / (sum of points in category c after drops). If you instead give every quiz an equal slice of the quiz bucket while quizzes have different point values (or one is dropped), the item-by-item result won’t match the category-first result. Different denominators (10-point quizzes vs 100-point projects) are fine if you normalize inside categories (use percentages within each category) before applying category weights. Dropping the lowest score does not change the advertised category weight in a category-weighted scheme, but it does redistribute that fixed bucket across the remaining items; in a points-based scheme it changes total points and can shift the effective weight of that category in the overall. So the “right” method is: follow the syllabus-category weights imply category-average-then-weight; total-points implies item-level point weighting. Without your exact rules I can’t be certain, but staying consistent with one scheme will remove the mismatch.

  2. Think of weighted averages like your smoothie: blending each fruit type first is fine, but when you mix the fruit-purees, each puree must be poured in proportion to how much actual fruit it represents. Translating to grades: if the syllabus says “quizzes 20%, projects 30%, exams 50%,” then the “right” path is to compute each category’s average according to that category’s rules (point-weighted vs equal-per-item, lowest dropped, etc.), and then multiply those category averages by 20%, 30%, 50% and add. If instead you toss every single item into one big blender and weight by raw points across the whole course, you’re using a different recipe where a category’s influence is determined by how many total points it happened to have-not the posted percentages-so you’ll generally taste a different final blend.

    Under the hood, there’s a tidy condition for when “average the categories then weight” equals “weight all items at once.” Let each item i carry an internal weight w_i (often its points, or 1 if each item is equally weighted in that category). The one-big-blender overall is sum(w_i s_i)/sum(w_i). If you group by category c, the category average A_c equals sum_{i in c}(w_i s_i)/sum_{i in c} w_i. These two methods match exactly if and only if the category weight you use equals W_c / sum W, where W_c = sum_{i in c} w_i. But most syllabi fix category weights (like “quizzes are 20% no matter how many points of quizzes exist”), which breaks that equality on purpose. Dropping a lowest score doesn’t change the category’s overall share in the category-weighted method (it just redistributes weight among the remaining items inside that category), but it will change the category’s effective share if you’re doing the all-items-by-points method-another reason the two routes disagree. So, unless your course explicitly says “final grade is total earned points over total possible points,” the correct approach is: follow each category’s internal policy to get its average, then apply the stated category percentages.

  3. Your two results differ because you’re mixing grading models: if the syllabus is category-weighted, first compute each category percent after any drops and then apply the fixed category weights; if it’s points-based, weight every item by its points over the total, and note that dropping a score changes a category’s effective weight only in the points-based model. Example: quizzes 10, 8, 6/10 (drop 6), projects 90, 70/100, exams 80, 90/100 → category-weighted: 0.2×90% + 0.3×80% + 0.5×85% = 84.5%, but points-based: (18+160+170)/(20+200+200) = 82.9%; see https://www.khanacademy.org/math/statistics-probability/summarizing-quantitative-data/weighted-averages/v/weighted-average-intuition.

Leave a Reply

Your email address will not be published. Required fields are marked *

Join Our Community

Ready to make maths more enjoyable, accessible, and fun? Join a friendly community where you can explore puzzles, ask questions, track your progress, and learn at your own pace.

By becoming a member, you unlock:

  • Access to all community puzzles
  • The Forum for asking and answering questions
  • Your personal dashboard with points & achievements
  • A supportive space built for every level of learner
  • New features and updates as the Hub grows