正解: ACorrect: A
所得分布は右に裾が長いため、平均値は少数の富裕層に引っ張られます。典型的な住民の購買力を知るには中央値や分布全体を見る必要があります。Income distributions are right-skewed, so the mean is inflated by a few very high earners. Median income and the full distribution better indicate typical purchasing power.
所得や資産のような分布は、上限がなく一部の極端な値が存在するため、平均値が中央値を大きく上回ることが普通です。Darrell Huff の古典『統計でウソをつく法』でも、平均という言葉が平均値・中央値・最頻値のどれを指すかで印象が大きく変わることが指摘されています。例えば住民の大半が中所得で、少数の富裕層が住む地域では、平均年収は高くても高価格帯サービスの顧客層は薄い可能性があります。選択肢Bは事実と逆で、右に歪んだ分布では平均が中央値より高くなります。Cは統計の全否定で建設的ではありません。Dは所得と需要を無条件に結びつけており、消費性向や競合状況を無視しています。Distributions such as income and wealth have no upper bound and contain extreme values, so the mean typically exceeds the median by a wide margin. Darrell Huff's classic How to Lie with Statistics notes how the impression changes depending on whether 'average' refers to the mean, median, or mode. A region where most residents earn middle incomes but a few are very wealthy may show a high average yet offer a thin customer base for premium services. Option B has the relationship backwards; in right-skewed distributions the mean is higher than the median. Option C dismisses statistics entirely, and Option D links income to demand unconditionally, ignoring spending propensity and competition.
中央値を見ても、所得が「二極化」した地域では中央値付近の住民が少ないという実態を捉えられません。また公的統計は世帯単位か個人単位か、税引前か後かで意味が変わり、指標の定義を確認せずに比較すると誤った結論に至ります。Even the median can mislead in a polarized region where few residents actually earn near the median. Public statistics also differ in whether they are household or individual, pre-tax or post-tax, and comparing them without checking definitions leads to wrong conclusions. Cost of living also varies, so nominal income alone overstates purchasing power in expensive areas.
平均値・中央値に加えて分位点（上位10%・25%など）やヒストグラムで分布の形を確認し、想定顧客層の人数を直接推計します。可能であれば既存の高価格帯サービスの利用実績や小規模なテスト販売で、統計から得た仮説を実際の購買行動で検証します。Look beyond the mean and median at quantiles, such as the top 10 percent and 25 percent, and histograms to see the shape of the distribution, then estimate the actual number of target customers. Where possible, validate the statistical hypothesis with data on existing premium services or a small test launch.